Chameth.com - posts like debugging-beyond-the-debugger, securing-all-the-things-with-1password but not migrating-from-github-to-forgejo, sense-api, why-you-should-be-using-https

Securing all the things with 1Password

2025-11-26T00:00:00Z

For many years I’ve been a keen user of Bitwarden. Recently I’ve had a lot of small paper-cut problems. The browser extension was redesigned and just doesn’t quite work how I expect any more. The prompt to save new login info misfired more than it worked. The mobile app stopped background refreshing properly. No one issue was enough to make me want to leave Bitwarden, but it definitely wasn’t sparking joy. Then there were rumblings about a price increase. I don’t think they’ve announced anything, but the numbers I saw were a jump from $10/year to $24/year. A nearly 2.5x price increase for a tool that was actively getting worse over time did not seem like a good deal to me¹.

I spent a while looking at alternatives, but there were very few that met my needs. Nordpass is highly praised but doesn’t support saving TOTP codes on personal plans². All the nice open-source/self-hosted solutions fall down when you want reliable synchronisation to an iOS device (which is fair enough). Proton Pass looked like it had all the features, but the pricing was asinine. Different pages showed different prices, the prices were introductory rates that jumped up on renewal, but that was hidden away in small print, etc. Yuck. The prices are also pretty high. More or less the same as 1Password, which I’d been discounting due to the cost, until now…

After eliminating basically every other option, I took a serious look at 1Password. It seemed to tick all the right boxes, I’d often heard good things about it, and it wouldn’t actually be much more expensive than Bitwarden if they went ahead with the rumoured price increase. I started a trial to give it a test.

Initial impressions

1Password is — as you’d expect for something that originated as a Mac app — very pretty. But it’s also highly functional. The looks don’t get in the way, they’re nice extras on top. When it’s locked, you get a large version of the 1Password logo, which is a front-on view of the barrel of a lock, next to the password input. When you authenticate, the barrel turns as though you’d put a key in and rotated it, and then the background splits in two and opens like a vault door. Does a password manager need to do that? No. Does it make me a little bit happy every time I see it? Yes. Look at it, it’s amazing!

The design philosophy carries through the rest of the app. Despite using Bitwarden daily, I often clicked on the wrong thing when I wanted to do anything other than autofill; in about a month of using 1Password I don’t think I’ve ever misclicked or been confused about the UI. The browser extension just feels snappier than Bitwarden, too. I’m not sure if it’s actually faster or if the design just makes it feel that way.

Functionality-wise, 1Password does everything Bitwarden did for me but slightly better. Bitwarden kicked you to its web interface for some operations (like checking compromised passwords); 1Password has that all built into the app. Instead of editing entries in a fiddly little browser popup, 1Password just summons forth the desktop app and you edit it there. It does a good job of detecting QR codes for TOTPs and offering to save them. I think the only small complaint I have is that the browser extension doesn’t have an autofill keybind. You can rig that up via the desktop app, but it feels like 1Password leans towards you using the UI it adds on login forms. To be fair, that UI works really well, and is very useful when you have multiple accounts on a site. My muscle memory for autofilling is strong, though.

Diving deeper

At this point I’d decided to switch fully to 1Password, so I started digging into what else it could do. The “Watchtower” feature combines a bunch of reporting about passwords (weak, breached, reused, and so on), but also lists sites where 1Password knows you can use passkeys or 2FA and don’t have them set up. I’d previously disabled passkey support in Bitwarden because it just appeared one day and got in the way of me trying to use a YubiKey, but decided to give them a go with 1Password. “Passkeys” is basically just a marketing term for FIDO2’s WebAuthn with automatic discovery bolted on top. The big problem is how you synchronise or export them. It’s all well and good letting your iPhone create a passkey when the Amazon app arbitrarily prompts it to, but what happens when you log in on a different device, or swap to Android, or whatever? Having 1Password deal with them fixes that³.

The passkey flow is very smooth: 1Password pops up an account selector, with obvious options to create a new account or pass the request through to another device like a YubiKey. It’s smart enough to guess the right account most times, even when you’ve got 17 different Google accounts for $reasons. Using them is equally simple: it just shows a popup and you tap to confirm you want to use the passkey. The Watchtower functionality made it really easy to go through and add passkeys everywhere that supports them. The thing I really appreciate is that they make the data that powers it public. It’s obviously partially a marketing thing for them, but it’s also a useful resource they could have kept proprietary.

Passkeys are a bit basic, though. 1Password has many more fun offerings. There’s an entire developer section with various goodies. At first I thought “that’s nice but I won’t use it”, but now I’m using basically everything it offers. The first thing to draw me in was the CLI. It’s pretty straightforward, but like the browser extensions it can nicely integrate with the desktop app so you unlock them all at once, deal with auth prompts in the GUI, and so forth. I have a couple of command-line tools that need passwords (e.g. for dealing with encrypted backups); now instead of prompting for the passwords, they shell out to op read and I authorise it via a popup in the 1Password app. Simple, but a nice quality of life improvement.

Next up there’s a beta feature called “Environments”. This lets you define groups of environment variables within 1Password, and have them exported automatically to certain destinations. At the minute it supports exporting to the AWS SecretsManager, and local .env files. When making a local file it actually makes a named pipe, so the credentials aren’t just sitting on disk (and won’t end up accidentally committed to anything). When the 1Password vault is locked, anything trying to read from the file will cause an authentication popup, much like with the CLI tool. I used to have the mentality that if something could read arbitrary files on disk, it was basically “game over” from a security point of view. Recent events made me realise there’s some value in defence-in-depth there, though, and 1Password provides a nice solution.

Fingerprints + Linux = Sadness

One feature of 1Password had me feeling a bit left out. Instead of entering your vault password to unlock it every time, you can have it invoke the system authentication library. On an iPhone this uses Face ID, on a Mac it can invoke Touch ID, and on Linux it uses Polkit. In the default configuration, Polkit just asks for your local user password, which is not much of an improvement. A fingerprint scanner seemed like the obvious solution here: it’s both a lot more convenient than typing a long password, and provides some assurance that it’s actually me doing the action⁴. Unfortunately, fingerprint readers on Linux seem to be a bit of a sorry affair.

Don’t get me wrong: there’s support. libfprint has a big list of supported devices, but I found it almost impossible to actually find a device I could buy that had one of the listed sensors. Most of them seem to be exclusively built into laptops. A lot of the cheaper devices just use whatever sensor they can get their hands on, so it’s pot luck if you receive a version that will work with libfprint or one that won’t. There’s also not a lot of information beyond that published by libfprint. I guess it makes sense: external fingerprint readers just aren’t a thing most people care about. While that makes me feel all special, it’s somewhat prohibitive to getting an actual working solution.

I’d given up on finding an answer to this when I stumbled upon an article by Scott Laird. He uses a YubiKey Bio to do it, via pam_u2f. The Bio isn’t actually a fingerprint reader: it just has an on-board reader to secure the credentials. The OS doesn’t get involved in the process. But because it can provide a fingerprint-secured U2F key, and pam_u2f can require that key to perform system operations, you basically get the same result. Maybe it’s actually even better than using a “real” fingerprint reader, as the data is secured on the hardware device…

The YubiKey Bio was a bit more pricey than I really wanted, coming in at around £90, but it felt like a worthy investment to add a bit of security and avoid typing a password dozens of times a day. The setup was simple: the official app that you use to enrol fingerprints is packaged in the Arch User Repository, and it worked perfectly. I already had pam_u2f set up as I used a (non-Bio) YubiKey to gate access to sudo, so I just generated a new key on the Bio, updated my config to use that, and added it to the Polkit policy. Now when I need to unlock 1Password or authenticate something, it prompts me to scan my fingerprint. Perfect!

Biometric all the things \o/

One of the other things Scott mentioned in that article was SSH keys. I’d seen 1Password supports acting as an SSH agent, but didn’t really see why I’d want it to. I used a resident key on my old YubiKey, so I didn’t have much to gain from a security point of view. Scott’s article made me realise I could have some big convenience gains, though. Currently every time I need to use the key, I enter a PIN and then physically touch it. I don’t mind this for the most part, but it turns out git-lfs does something like four different requests every time you do a git operation, none of which reuse the same SSH connection for some reason. That gets painful when each request involves a PIN and a touch. I also liked the idea of using pam_rssh for sudo authentication, but didn’t want to have to do the PIN-and-touch dance every time I used it.

You can probably see where this is going: I generated an SSH key in 1Password, and then… spent literal hours disabling all the other SSH agents that were running on my system, and trying to figure out where SSH_AUTH_SOCK was being populated. I didn’t manage to figure that out in the end⁵, and ended up just overriding it to point at 1Password’s agent socket. I added the new key to GitHub, and did the usual ssh git@github.com test, and it worked perfectly. 1Password prompts to allow access to the key, showing the process requesting access, and authorising it requires a fingerprint read:

can haz ssh key?

I’m not sure why Kitty doesn’t get an icon in the prompt; other applications do. You can expand the details section to see the key fingerprint being requested, as well as the exact process name and PID requesting the key. With the proof of concept done, I generated a couple more keys: one for normal SSH, and one for gaining root access using pam_rssh. Using different keys ensures I get prompted for them separately, and means a malicious app couldn’t sneakily ask for a Git key but actually pivot to SSH into a server. Not sure that’s ever likely to be a problem, but it’s basically no additional work for a bit more defence-in-depth, so why not?

I also took a moment to configure pam_u2f for both lightdm and xfce4-screensaver, so I can now log in and unlock my PC with my fingerprint. It’s not quite as smooth as Touch ID on a Mac, but it’s pretty close.

As an aside: this is not the first time I’ve been in this situation. Both Spotify and Todoist have focused on features I don’t want or need (including, but not limited to, pretty much everything branded “AI”), and then hiked the price up afterwards. It’s not quite standard enshittification, but it’s clear that our incentives no longer align properly. ↩︎
Or didn’t, at the time I looked. It now says “All plans” on the website but the docs only exist for business accounts, and there doesn’t seem to be any announcement explaining the change. ↩︎
Having Bitwarden deal with them would also have fixed that, but it got in my way instead of helping me, so… ↩︎
or it’s not me, but I’ll be busy dealing with the presumably-undesired removal of one of my digits, so probably won’t care about whatever is happening on the computer. ↩︎
Although I’m pretty sure the answer starts and ends with “systemd, somehow”. ↩︎

How I Get Things Done

2024-09-01T00:00:00Z

I always quite like reading about how other people do things. What software or hardware they use, or how they manage reminders, todo lists, and so on. I’ve never actually written about how I do any of that, though. So here it is!

Productivity

In the past I’ve fallen victim to the idea of there being One True Productivity System that would solve all my problems and make me amazing at getting things done. The title of this post is a nod to Getting Things Done, which I’ve read and tried to religiously follow in the past, but it’s just not for me. One thing that did actually stick from it, though, was the idea of “open loops”:

Anything that does not belong where it is, the way it is, is an “open loop,” which will be pulling on your attention if it’s not appropriately managed.

If I want to do something and it’s not recorded in a way I trust, it weighs on me a little. Those little weights all add up, and just make life uncomfortable. I deal with those in a few ways:

Inbox Zero-ish: if an e-mail needs me to do something, it sits in my inbox until it’s done. When it’s done, it gets archived.
Notes: random things I try to remember I put into my notes, so I can search for them later.
Budget: anything monetary I just adjust in my budget or create a category for.
Todo list: any other kind of task I want/need to do either now or at some point in the future goes into Todoist.

Todoist

In the past I’ve gone through a lot of different todo systems, once again hunting for one that will fix all my problems etc¹. In the end I always ended up coming back to Todoist, and have stuck with it for a decent period now.

In Todoist, I have everything split up into projects. Work projects are one colour and personal projects are another. Anything that needs more than a handful of individual tasks gets a project, which is archived when it’s done. Any one-off things that don’t neatly fall into projects go into “Random junk” project.

As I’m writing this, I have 116 items in 22 projects. A good deal of those are things that I might want to do at some point, not things that I actively need to worry about. If I want to get one of them done, I schedule it so it shows up in the “today” view. I have the Todoist desktop app open automatically when I login to my computer, and it gets positioned on the right hand side of my ultra-wide monitor, so I can see the “today” view and add/remove things immediately at any point.

I’m not precious about tasks becoming overdue, or unscheduling things I’d previously scheduled. Priorities change, ideas that sounded good a week ago might no longer sound good, and so on.

I do a few things to help me manage tasks and keep track of things:

I have a monthly recurring task to look through everything that’s not scheduled, just so I keep a rough idea of what’s there, and can update or get rid of it as needed.
I wrote todoistager and have it running on a server to automatically apply tags to tasks based on their rough age: weeks, months or years. This lets me identify things that are lingering and reconsider them².
I automate adding tasks for things that demand my attention, like incoming GitHub pull requests. The Todoist API is nice and straight forward, and hooking it up to things is trivial.

Other desktop software

For notes I use Logseq, and I just put whatever I need to write under today’s note and rely on search or tags to find it again when I need to. I note basically anything that I might like to use again in the future. One of my most common tags is #blub, a term I came across in the excellent article “In defense of blub studies”. Blub is random ultra-specific knowledge; it might not be eminently useful to know on its own, but over time it can pay dividends.

I’ve previously tried a bunch of note taking apps, and methodologies, but like with todo lists I end up just wanting something that just works and stays out of the way. Logseq stores notes in Markdown, and I have a cronjob to commit and push my changes to a private git repository.

Like Todoist, Logseq launches when I log into my computer, and sits in the same 500px-or-so region on the right of my monitor. This allows me to switch from Todoist to take or refer to notes when needed. To do this positioning, I use devilspie2. I also use this to position Discord and my IRC client side-by-side on a small portable monitor that sits underneath my ultrawide.

More mundane software: I use IntejjiJ IDEA for almost all development and complex text editing³. The only exception is Android development where Android Studio rules the roost. For lighter text editing, Sublime Text is my choice.

I spend a fair bit of the time at the command line. I use kitty as a terminal emulator, with the zsh shell and Oh My Zsh for some nice plugins and enhancements. One thing I couldn’t live without is autojump, which allows quickly jumping to any directory you’ve previously visited using a substring of the name.

Web browsing is all Firefox, with Bitwarden for password management and two pinned Fastmail tabs, one for personal e-mail and one for work. I don’t think there’s much else to note there.

Oh, I use Arch, by the way.

Infrastructure

This is the hodge-podge of things that support my computer usage but sit outside the actual computer. In no particular order…

All my DNS queries go through NextDNS using DNS-over-HTTPS. NextDNS lets you add custom responses, block certain things from resolving, and subscribe to pre-made anti-tracking or anti-advertising lists. It gives you far more control over everything than most other services I’ve looked at.

Every device I control runs Tailscale to enable remote access. I also use Tailscale’s SSH authentication feature for my servers, allowing quick and easy access from my phone where I don’t have or want normal SSH keys.

One exciting service that combines these two is golink. It was created by Tailscale, presumably based on the go service used extensively inside Google. I run a version on my tailnet, and have NextDNS configured to resolve go to the tailscale IP of the service. That lets me type “go/whatever” on any machine I set up with NextDNS and Tailscale, including mobile devices. I use go links for all sorts of things:

Shortcuts to tools I use commonly: go/board gets updated to my current client’s kanban board or similar tracker, whether it’s in Jira, GitHub, Asana or whatever; go/meet goes to Google Meet with the user parameter set so I’m logged in with my work account not my personal account.
In place of bookmarks: go/flex goes to the CSS-Tricks flexbox guide⁴, go/cad goes to the OnShape login page, go/hmrc goes to the HMRC online services page, and so on
A few dynamic things: go/github/chameth.com will go to the csmith/chameth.com project on GitHub; go/ref/pipico will go to the Pi Pico page on my electronics reference wiki.

For my servers I have a minimal Ansible setup that does some basic configuration and deploys SSH keys to the right places. Almost everything runs inside Docker, and I use my own software for HTTP proxying: Dotege uses tags I apply to containers to build a config for Centauri which handles reverse proxying, using Let’s Encrypt for TLS certs. At the minute I have two servers: a dedicated server that runs basically all my services (including this website), and a little VPS for monitoring and backups. Both are from Hetzner.

I wrote my own basic monitoring software called Goplum to make sure things that should be running on my servers actually are. If it fails checks it sends a notification to my phone using Pushover, and a message to a channel on my private IRC server (running Ergo).

Hardware

For the last two years I’ve used a laptop as my main computer⁵. It’s a Dell G15 Special Edition. After some upgrades its specs are:

Intel i7-12700H processor
NVIDIA GeForce RTX 3060 graphics card
2x32GB DDR5 RAM
2TB Gen4 SSD

At home it sits behind my Alienware AW3423DWF monitor. This is a Quantum Dot OLED monitor, and I bought it almost immediately upon seeing someone else use it at a LAN event. As did three other people that were sat near us. It just looks so good. I wasn’t sure I’d get on with a widescreen monitor, and made sure it supported picture-by-picture before buying it, but have never actually used that. I generally run whatever I’m working on at nearly-fullscreen, then have Todoist or Logseq on the far right. Games go fullscreen over everything, of course.

Above the widescreen I have a curved lightbar that I find helps reduce eyestrain and makes it easier to focus on the monitor. And on top of that I have a Logitech Brio webcam which gives a great picture with lots of control over pan/zoom/etc for meetings. I have a little set of systemd scripts that detect when the external webcam is connected and disable the laptop’s built-in webcam, and vice-versa when it’s disconnected; that means whenever I jump into a video call in any app there’s only one input and it’s always the right one, which is nice.

Below the main monitor and its friends, I have a generic 1080p portable monitor connected to the laptop via USB-C. This comes with me when I travel, and serves as a Discord/IRC monitor when at home. It’s handy for throwing other things on as well, like if I need to reference a design while writing code, and don’t want to keep alt+tabbing.

I have two Durgod K320 keyboards: one with Cherry MX Silent Red switches for when I’m at home and don’t want to annoy people with the sounds, and one with Cherry MX Blues for when I’m travelling and can make a bit more noise in exchange for a much nicer typing experience. They’re both tenkeyless so they can handily fit in a backpack.

I use a Zowie EC1-C wired mouse, which does the job well. I got fed up with mice from more mainstream manufacturers failing: my previous mouse was a Razer Viper that lasted 7 months before it started losing clicks. All my audio goes through a HyperX Cloud Alpha Wireless, which has the most insanely good battery life of any wireless headset I’ve ever owned or seen. Their claim of 300 hours is not just marketing!

In non-computer hardware, I have a Bambu P1S 3D printer that I’ve mentioned in my blog before. For printing in fewer dimensions I have an Epson EcoTank ET-1810, which is surprisingly reasonable for an inkjet printer, and the ink costs aren’t as astronomical as normal cartridges. Plus, you can’t decide to DRM ink when it’s loaded raw into the printer…

I have a small Zigbee network set up mainly to control the lights and the old 3D printer I no longer use (the P1S has a sensible standby mode so you don’t have to yank its power away to make the fans stop spinning!). That’s run through a Raspberry Pi running Zigbee2Mqtt and a custom app I wrote to monitor events on MQTT and generate the appropriate responses (or just log the data).

My daily-driver phone is an iPhone 15 Pro, but I have a bunch of Android devices kicking around for work purposes. I read nearly every day on a Kindle Paperwhite Signature Edition.

Money and work

I use You Need A Budget for figuring out how much money I can spend on what, and making sure I have enough put aside for bills and taxes and so on. I use it for both business and personal accounts together, but I mark all the business categories and accounts with a 💼 emoji, and have some scripts that use the API to make sure the money split makes sense⁶.

Every recurring monthly and annual payment has a category, as do all upcoming events I plan to attend, and all the wonderful taxes I have to pay. I also have categories for hobbies and other expenditure. I think all of that is pretty typical for a normal YNAB setup.

I also use YNAB for tracking anything I might like to buy at some point: it gets a category in a “Wishlist” group, and I’ll occasionally budget to buy things from there. I try (and often fail) to use that as a way to put off impulse purchasing. As my income is often very inconsistent, I try to set aside money so that I can “spend” a bit every week on things in the wishlist group. That feels a lot nicer to me than just budgeting everything in one go as soon as I get paid, and gives me more time to consider whether I really want those things⁷.

For work, I do all the more technical accountancy and tax filing and so on through FreeAgent. For my fairly straight-forward tax situation, I can basically manage everything myself using FreeAgent without paying for an accountant. That does mean occasionally spending a bunch of time reading tax manuals, but I’d rather fully understand the rules than just follow advice anyway.

The problem with implementing productivity systems or switching todo apps or anything of that ilk is that it can feel like you’re being productive by doing so. It takes a while to realise that you’re not, and actually getting on with things is better. ↩︎
Or just accept that I’ll maybe do them at some point. I kept the labels vague to try and not make it feel judgmental. ↩︎
Including writing this blog post right now. ↩︎
One day I will get all the justify- and align- options right without looking them up, but until then… ↩︎
This is very normal for developers, but much less common for computer gamers. I’ll probably go back to a desktop next time I upgrade, as I’m travelling a lot less. ↩︎
I can’t just arbitrarily move money from the business to myself, it has to be given as salary or a divided etc, which has tax implications. If I budget “business money” for personal things, then it actually has to be properly transferred and accounted for before being spent (as I’m in effect budgeting a future paycheck). It would be cleaner to keep the accounts completely separate, but I find this way works well for me as I can be a bit more flexible. ↩︎
It feels a little like I’m giving myself pocket money, which is a bit weird, but it’s a lot better than having no “income” for several months when client invoices fall weirdly or I’m between contracts. ↩︎

Debugging beyond the debugger

2019-05-08T00:00:00Z

Real-life debugging tools

Most programming — and sysadmin — problems can be debugged in a fairly straight forward manner using logs, print statements, educated guesses, or an actual debugger. Sometimes, though, the problem is more elusive. There’s a wider box of tricks that can be employed in these cases but I’ve not managed to find a nice overview of them, so here’s mine. I’m mainly focusing on Linux and similar systems, but there tend to be alternatives available for other Operating Systems or VMs if you seek them out.

Networking

tcpdump

tcpdump prints out descriptions of packets on a network interface. You can apply filters to limit which packets are displayed, chose to dump the entire content of the packet, and so forth.

Typical usage might look something like:

# tcpdump -nSi eth0 port 80
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on eth0, link-type EN10MB (Ethernet), capture size 262144 bytes
16:03:35.577781 IP6 2001:db8::1.54742 > 2001:db8::2.80: Flags [S], seq 2815779044, win 64800, options [mss 1440,sackOK,TS val 2378811665 ecr 0,nop,wscale 7], length 0
16:03:35.586853 IP6 2001:db8::2.80 > 2001:db8::1.54742: Flags [S.], seq 1522609102, ack 2815779045, win 28560, options [mss 1440,sackOK,TS val 3063610173 ecr 2378811665,nop,wscale 7], length 0
16:03:35.586877 IP6 2001:db8::1.54742 > 2001:db8::2.80: Flags [.], ack 1522609103, win 507, options [nop,nop,TS val 2378811674 ecr 3063610173], length 0
16:03:35.620678 IP6 2001:db8::1.54742 > 2001:db8::2.80: Flags [P.], seq 2815779045:2815779399, ack 1522609103, win 507, options [nop,nop,TS val 2378811708 ecr 3063610173], length 354: HTTP: GET / HTTP/1.1

Here you can see the start of a plaintext HTTP request: the three-way handshake as the TCP connection is established followed by a GET request. Even if the data is encrypted as it will be in most cases, it’s often useful to see the “shape” of the transmissions: did the client start sending data when it connected, did the server ever respond, etc.

Daniel Miessler has a good tutorial on tcpdump if you’re not familiar with it and don’t want to jump straight into the man page.

… with Docker

Docker sets up separate network namespaces for each container. To see the traffic across the interfaces of a single container you can nsenter the container’s network namespace:

# nsenter -t $(docker inspect --format '{{.State.Pid}}' my_container) -n tcpdump -nS port 80

This retrieves the PID for the container, and tells nsenter to enter the network (-n) namespace from the given target (-t) PID, and then run the given command (in this case tcpdump ...).

openssl s_client / s_server

When a connection is using TLS it’s often useful to try connecting to the server and see what certificate it presents, algorithms it negotiates, and so forth. OpenSSL offers two useful subcommands which can help with this: s_client for connecting as a client, and s_server for listening to connections.

For example, using s_client to connect to google.com on the standard HTTPS port shows us details about the server cert and its verification status:

$ openssl s_client -connect google.com:443
CONNECTED(00000003)
depth=2 OU = GlobalSign Root CA - R2, O = GlobalSign, CN = GlobalSign
verify return:1
depth=1 C = US, O = Google Trust Services, CN = Google Internet Authority G3
verify return:1
depth=0 C = US, ST = California, L = Mountain View, O = Google LLC, CN = *.google.com
verify return:1
---
Certificate chain
 0 s:C = US, ST = California, L = Mountain View, O = Google LLC, CN = *.google.com
   i:C = US, O = Google Trust Services, CN = Google Internet Authority G3
 1 s:C = US, O = Google Trust Services, CN = Google Internet Authority G3
   i:OU = GlobalSign Root CA - R2, O = GlobalSign, CN = GlobalSign
---
# ...

Whereas connecting to my webserver and providing an unknown host in the SNI field results in an SSL alert 112 (“The server name sent was not recognized”) and no server certificate is sent:

$ openssl s_client -connect chameth.com:443 -servername example.com
CONNECTED(00000003)
140384831313024:error:14094458:SSL routines:ssl3_read_bytes:tlsv1 unrecognized name:../ssl/record/rec_layer_s3.c:1536:SSL alert number 112
---
no peer certificate available
---
# ...

Often if you hit this kind of alert in an application the exact error will be lost somewhere in the many layers between the SSL library and the logs, so being able to directly connect and test can help diagnose a lot of issues.

Once a connection is established you can read and write plain text and it will be encrypted and decrypted automatically.

Java apps

If a Java app is involved in the connection, you can enable a lot of built-in debugging with a simple JVM property: javax.net.debug. You can tweak what exactly gets logged, but the easiest thing to do is just set the property to all and you’ll see information about certificate chains, verification, and packet dumps:

$ java -Djavax.net.debug=all -jar ....
# ...
found key for : duke
chain [0] = [
[
  Version: V1
  Subject: CN=Duke, OU=Java Software, O="Sun Microsystems, Inc.",
  L=Cupertino, ST=CA, C=US
# ...

More information about Java’s debugging options is available on docs.oracle.com.

Thread and core dumps

Higher-level languages frequently provide an interactive way to dump the current execution state of all of their threads (a “thread dump”). This is useful to spot deadlocks, some types of race conditions, and as a quick and dirty method of investigating hangs or excessive CPU usage.

With both Java and Go applications you can send a QUIT signal to have a thread dump printed out; Go applications will quit after doing so, Java ones will carry on running. At most terminals you can hit Ctrl and \ to send a QUIT signal.

For Java you can also use the jstack tool from the JDK to dump threads by PID; this can be useful if the application is running in the background or has redirected sysout:

$ jstack 8321
Attaching to process ID 8321, please wait...
Debugger attached successfully.
Client compiler detected.

Thread t@5: (state = BLOCKED)
 - java.lang.Object.wait(long) @bci=-1107318896 (Interpreted frame)
 - java.lang.Object.wait(long) @bci=0 (Interpreted frame)
 - java.lang.ref.ReferenceQueue.remove(long) @bci=44, line=116 (Interpreted frame)
 - java.lang.ref.ReferenceQueue.remove() @bci=2, line=132 (Interpreted frame)
 - java.lang.ref.Finalizer$FinalizerThread.run() @bci=3, line=159 (Interpreted frame)

# ...

A core dump provides more complete information about the state of a process, but is often more complex to interpret. The gcore utility from GDB will create a core dump of a process with a given PID. You can then generally load the core file using your normal debugger, depending on the language in question.

System calls

strace is the swiss army knife for seeing what a process is doing. It details each system call made by a program (you can filter them down, of course). For example:

$ strace -e read curl https://google.com/
read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0 \236\0\0\0\0\0\0"..., 832) = 832
read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0P!\0\0\0\0\0\0"..., 832) = 832
read(3, "\177ELF\2\1\1\3\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\200l\2\0\0\0\0\0"..., 832) = 832
read(3, "\177ELF\2\1\1\0\0\0\0\0\0\0\0\0\3\0>\0\1\0\0\0\20Q\0\0\0\0\0\0"..., 832) = 832
# ...
read(3, "\0\0\0\0\0\0\0\4\25\345\366\302\273sE6\365wI\225\321|\3435Z\362\216\372\215\251aO"..., 253) = 253
<HTML><HEAD><meta http-equiv="content-type" content="text/html;charset=utf-8">
<TITLE>301 Moved</TITLE></HEAD><BODY>
<H1>301 Moved</H1>
The document has moved
<A HREF="https://www.google.com/">here</A>.
</BODY></HTML>
read(3, "\27\3\3\0!", 5)                = 5
# ...

Brendan Gregg has a nice guide on strace and alternatives.

… with docker

When the application is running in docker you can usually just strace it from the host with the correct PID (from e.g. docker inspect --format '{{.State.Pid}}' my_container). Sometimes you may need to trace the startup of an application though, which is a bit trickier. Instead you can run a new container using the same PID namespace as your target, and the permissions needed to strace:

$ docker run --rm -it --pid=container:my_container \
  --net=container:my_container \
  --cap-add sys_admin \
  --cap-add sys_ptrace \
  alpine

From within the new container you can install strace, and trace any running program within the target container using strace -p as normal. To start a new program you need access to the target container’s filesystem, which you can get to via /proc/1/root (PID 1 being the main process that docker started in the target container).

Files

Sometimes the problem might relate to file access. There are a couple of straight forward — but nonetheless useful — tools which might help here. inotifywait uses the Linux inotify subsystem to watch files or directories for operations. For example:

$ inotifywait -mr site/content
Setting up watches.  Beware: since -r was given, this may take a while!
Watches established.
site/content/post/ MODIFY 2019-05-08-debugging-beyond-the-debugger.md
site/content/post/ OPEN 2019-05-08-debugging-beyond-the-debugger.md
site/content/post/ MODIFY 2019-05-08-debugging-beyond-the-debugger.md
site/content/post/ MODIFY 2019-05-08-debugging-beyond-the-debugger.md
site/content/post/ CLOSE_WRITE,CLOSE 2019-05-08-debugging-beyond-the-debugger.md
# ...

Here the -m switch makes inotifywait monitor the files forever (instead of exiting on the first modification, which is the normal behaviour) and r makes it recurse into the directory and monitor each file and subdirectory in there.

If you want to see what processes currently have a file open, fuser is the go-to tool. For example:

$ fuser -v /
                     USER PID ACCESS COMMAND
/:                   root     kernel mount /
                     chris      2961 .rc.. systemd
                     chris      2986 .r... gdm-x-session
                     chris      2994 .r... dbus-daemon
                     chris      3001 .r... gnome-session-b
# ...

Honourable mentions

These aren’t really debugging tools, but I feel it’s worth mentioning as they often feature somewhere along the debugging-of-weird-problems journey.

I’ve seen some weird and wonderful problems happen because a disk is full, so a quick df early on in the debugging process never hurts. Some apps may hang, some may corrupt their config, some may fall over and die; sometimes the manner in which they fail doesn’t obviously point to a disk space issue.

Another issue that comes up now and then — especially inside VMs or other environment that don’t have a decent amount of “noise” happening — is entropy exhaustion. A quick look at /proc/sys/kernel/random/entropy_avail should be enough to confirm that everything is ticking along nicely. If it’s exceedingly low then you may find that anything involving random number generation stalls (TLS connections for example).