Commit graph

7 commits

Author SHA1 Message Date
Florian Klink
6b6a34065e feat(tvix/tracing): add tracing-tracy support
This introduces another feature flag, "tracy" to the `tvix-tracing` crate.

If enabled (not enabled by default), it'll add an additional layer
emitting packets in a format that https://github.com/wolfpld/tracy can
display.

I had to be a bit tricky with the combinatorial complexity when adding
this, but the resulting code still seems manageable.

Change-Id: Ica824496728fa276ceae3f7a9754be0166e6558f
Reviewed-on: https://cl.tvl.fyi/c/depot/+/10952
Tested-by: BuildkiteCI
Reviewed-by: Simon Hauser <simon.hauser@helsinki-systems.de>
Reviewed-by: flokli <flokli@flokli.de>
2024-06-14 19:33:44 +00:00
Florian Klink
d25f047b9d refactor(tvix/tracing): move otlp setup into helper function
Having all this in the main control flow makes it a bit hard to read.
Moving it into a helper function makes it a bit cleaner.

Change-Id: Ibdb739dbd1e013b4f8c4aaf9b036a6bd556a1871
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11814
Autosubmit: flokli <flokli@flokli.de>
Reviewed-by: Simon Hauser <simon.hauser@helsinki-systems.de>
Tested-by: BuildkiteCI
2024-06-14 14:52:19 +00:00
Simon Hauser
118e69d81d fix(tvix/tracing): reduce the error logs of otlp if collector is offline
The problem is that opentelemetry_otlp tonic batch exporter tries to
exports if either the `scheduled_delay` or if the
`max_export_batch_size` is reached. Per default the
`max_export_batch_size` is set to 512 spans, which means that we try to
export these spans once that counter is reached. Each export will then
try to connect to the exporter (if that not already happening) and will
result in a `tcp connect error`.
Increasing the max_export_batch_size to 4096 will then ensure that the
export only happens if the `scheduled_delay` is met after the 10
seconds.
`max_queue_size` is also increased, because `max_export_batch_size`
should not be greater than `max_queue_size`, so similar to the default
config its set to `max_export_batch_size * 4`.
This will reduce the amount of tries to otlp if the collector is not
available and otlp enabled.

Change-Id: Ic3430006e8a104fa3b34d274678cae55b3620ce9
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11791
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
Autosubmit: Simon Hauser <simon.hauser@helsinki-systems.de>
2024-06-14 10:20:31 +00:00
Simon Hauser
a857a2b978 feat(tvix/tracing): apply EnvFilter to all layers
Currently we apply the EnvFilter only to the stderr output writer.
This didn't affect any other layer, like the otlp layer, causing spans
from `h2`, `tokio_util` or other third party crate dependencies to be
always sent out via OTLP.

This changes that behaviour, applying EnvFilter to all exports, leading
to a lot less spans being exported.

Change-Id: I9f3a7233e9d0aeaa81fe08914579f0b3c80d134e
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11813
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
Autosubmit: Simon Hauser <simon.hauser@helsinki-systems.de>
2024-06-14 10:18:59 +00:00
Simon Hauser
fa7ed39bf4 feat(tvix/tracing): correctly close otlp on exit
Provide a new interface for forcing a flush of otlp traces and use this
interface to shutdown otlp prior to exiting tvix-store, either if the
tool was stopped with a SIGTERM or ended regularly.
This also fixes an issue where traces were not even exported if for
example we just imported 10 paths and never even emitted more than 256
traces. The implementation uses a mpsc channel so a flush can be done
without having to wait for it to complete. If you want to wait for a
flush to complete you can provide a oneshot channel which will receive a
message once flushing is complete.

Because of a otlp bug `force_flush` as well as
`shutdown_tracer_provider` need to be executed using `spawn_blocking`
otherwise the function will deadlock. See
https://github.com/open-telemetry/opentelemetry-rust/issues/1395#issuecomment-1953280335

Change-Id: I0a828391adfb1f72dc8305f62ced8cba0515847c
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11803
Reviewed-by: flokli <flokli@flokli.de>
Tested-by: BuildkiteCI
Autosubmit: Simon Hauser <simon.hauser@helsinki-systems.de>
2024-06-14 09:34:51 +00:00
Florian Klink
79bfa931ed feat(tvix/tracing): set release_max_level_debug for tracing
This allows explicitly opting in to get DEBUG-level log lines, by
setting RUST_LOG.

It currently also causes traces to be emitted in all cases, so we might
do some runtime filtering there too, as discussed in cl/11791.

Change-Id: I2865bb06a62465836d63196422f5f734f7165386
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11801
Tested-by: BuildkiteCI
Reviewed-by: aspen <root@gws.fyi>
Autosubmit: flokli <flokli@flokli.de>
2024-06-13 16:18:47 +00:00
Simon Hauser
825d498908 feat(tvix/tracing): introduce common tvix-tracing crate
Introduce a new common crate that contains tracing boilerplate which then
can be used in the cli, tvix-store and tvix-build crates.
It has otlp as an optional feature, which is currently only used by
tvix-store.

Change-Id: I41468ac4d9c65174515d721513b96fea463d6ed2
Reviewed-on: https://cl.tvl.fyi/c/depot/+/11758
Tested-by: BuildkiteCI
Reviewed-by: flokli <flokli@flokli.de>
Autosubmit: Simon Hauser <simon.hauser@helsinki-systems.de>
2024-06-10 16:35:08 +00:00