Since ca1dc3f70bf98e2424b7b2666ee2180675b67451, the NAR parser has moved
the preallocate & receive steps into the file handle class to remove the
assumption that only one file can be handled at a time.
The third argument to `open()` in `-|` mode is passed to a shell if it's
a string. In my case the store URI contains
`?secret-key=${signingKey.directory}/secret&compression=zstd`
For the `nix store cat` case this means that
* until `&` the process will be started in the background. This fails
immediately because no path to cat is specified.
* `compression=zstd` is a variable assignment
* the `$path` argument to `store cat` is attempted to be executed as
another command
Passing just the list solves the problem.
`dev-notes` are severely outdated. I dropped everything except one note
that I moved to hacking.md. The parts about creating users are also
covered elsewhere.
The `update-dbix` part got a just command to make it discoverable again.
* justfile, inspired from Lix.
* let foreman use the stuff from outputs, similar to what Lix does>
* mess around with PERL5LIB[1] and PATH to get tests running locally.
[1] I don't really know how `Setup` was found before tbh.
When an artifact is requested from hydra the output is first copied
from the nix store into memory and then sent as a response, delaying
the download and taking up significant amounts of memory.
As reported in https://github.com/NixOS/hydra/issues/1357
Instead of calling a command and blocking while reading in the entire
output, this adds read_into_socket(). the function takes a
command, starting a subprocess with that command, returning a file
descriptor attached to stdout.
This file descriptor is then by responsebuilder of Catalyst to steam
the output directly
The primary host for this repo isn't github, and this is causing a whole
bunch of false indications in the Lix Forgejo UI as it tries to run the
.github/workflow with non-existing runners.
Only log issues/failures when something's actually up.
It has irked me for a long time that so much output came
out of running the tests, this seems to silence it.
It does hide some warnings, but I think it makes the output
so much more readable that it's worth the tradeoff.
Helps for highly parallel running of jobs, sometimes they'd not give output for a while.
Setting this timeout higher appears to help.
Not completely sure if this is the right place to do it, but it works fine for me.
We've seen many fails on ofborg, at lot of them ultimately appear to come down to
a timeout being hit, resulting in something like this:
Failure executing slapadd -F /<path>/slap.d -b dc=example -l /<path>/load.ldif.
Hopefully this resolves it for most cases.
I've done some endurance testing and this helps a lot.
some other commands also regularly time-out with high load:
- hydra-init
- hydra-create-user
- nix-store --delete
This should address most issues with tests randomly failing.
Used the following script for endurance testing:
```
import os
import subprocess
run_counter = 0
fail_counter = 0
while True:
try:
run_counter += 1
print(f"Starting run {run_counter}")
env = os.environ
env["YATH_JOB_COUNT"] = "20"
result = subprocess.run(["perl", "t/test.pl"], env=env)
if (result.returncode != 0):
fail_counter += 1
print(f"Finish run {run_counter}, total fail count: {fail_counter}")
except KeyboardInterrupt:
print(f"Finished {run_counter} runs with {fail_counter} fails")
break
```
In case someone else wants to do it on their system :).
Note that YATH_JOB_COUNT may need to be changed loosely based on your
cores.
I only have 4 cores (8 threads), so for others higher numbers might
yield better results in hashing out unstable tests.
Lost in the h-e-j -> n-e-j migration, causing evaluation to always be
single threaded and limited to 4GiB RAM. Follow the config settings like
h-e-j used to do (via C++ code).
There are some known regressions regarding local testing setups - since
everything was kinda half written with the expectation that build dir =
source dir (which should not be true anymore). But everything builds and
the test suite runs fine, after several hours spent debugging random
crashes in libpqxx with MALLOC_PERTURB_...
The current way this whole build works is incompatible with having a
separate build dir, or at least with having a separate build dir. To be
improved in the future - maybe minimize the dependencies a bit. But this
isn't so much data that we really have to care.
New jobs have their "new" status take precedence over them being
"failed" or "queued", which means actions that can act on "failed" or
"queued" jobs weren't shown to the user when they could only act on
"new" jobs.
- Use extra-trusted-users to avoid overriding the default set of trusted
users and causing permission issues.
- Add hydra and hydra-www users which also need permissions.
nix-eval-jobs streams output, unlike hydra-eval-jobs. Now that we've
migrated, we can use this to:
1. Use less RAM by avoiding buffering a whole eval's worth of metadata
into a Perl string and an array of JSON objects.
2. Make evals latency a bit lower by allowing the queue runner to start
ingesting builds faster.
The feature cannot easily be ported to nix-eval-jobs since it requires
deep integration into the evaluator, and h.n.o doesn't use it. Later
more of this will be ripped out.
