feat(rel): add syntactic code intel worker service

[jdpleiness]

Add SCIW service to deployment

Checklist

• Sister deploy-sourcegraph change:
• Sister customer-replica change (if necessary, for any changes affecting pure-docker or configuration):
• All images have a valid tag and SHA256 sum

Test plan

TODO

[jdpleiness]

• feat(rel): add syntactic code intel worker service #1163 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

[jdpleiness]

@antonsviridov-src , trying to spin this up locally in testing, I'm seeing the following error:

{"SeverityText":"ERROR","Timestamp":1739907970056237468,"InstrumentationScope":"conf.client","Caller":"conf/client.go:305","Function":"github.com/sourcegraph/sourcegraph/internal/conf.(*client).continuouslyUpdate","Body":"received error during background config update","Resource":{"service.name":"syntactic-code-intel-worker","service.version":"6.0.0","service.instance.id":"b7c572526148"},"Attributes":{"triggered_by":"waitForSleep","error":"unable to fetch new configuration: rpc error: code = Unavailable desc = connection error: desc = \"error reading server preface: http2: frame too large\""}}

Any ideas what might be the cause here? 🙏

[antonsviridov-src]

@jdpleiness responded on slack, but it case you haven't seen:

So this seems to be a message by config machinery when frontend is unreachable: https://github.com/sourcegraph/sourcegraph/blob/5c4442eddd7d46e9d1804c61bf399c5ac4f651a6/internal/conf/client.go#L305

https://sourcegraph.slack.com/archives/C02E4HE42BX/p1740132375444449?thread_ts=1738238003.778069&cid=C02E4HE42BX

		// This needs to be long enough to allow the frontend to fully migrate the PostgreSQL
		// database in most cases, to avoid log spam when running sourcegraph/server for the
		// first time.
		delayBeforeUnreachableLog: 15 * time.Second,

which is by default 15 seconds, I've seen it take much much longer to start on my machine

[jdpleiness]

Ahh okay, I'll circle back to this and take a look. Thanks for the hunting down the issue 🙏

[jdpleiness]

@jdpleiness responded on slack, but it case you haven't seen:

So this seems to be a message by config machinery when frontend is unreachable: https://github.com/sourcegraph/sourcegraph/blob/5c4442eddd7d46e9d1804c61bf399c5ac4f651a6/internal/conf/client.go#L305

https://sourcegraph.slack.com/archives/C02E4HE42BX/p1740132375444449?thread_ts=1738238003.778069&cid=C02E4HE42BX

		// This needs to be long enough to allow the frontend to fully migrate the PostgreSQL
		// database in most cases, to avoid log spam when running sourcegraph/server for the
		// first time.
		delayBeforeUnreachableLog: 15 * time.Second,

which is by default 15 seconds, I've seen it take much much longer to start on my machine

@antonsviridov-src I updated the env vars for the service to match precise code intel like the other deployments. I'm still not able to get this healthy though. I also tried adding a depends_on clause for frontend to be sure frontend was up before this service started, but it still fails it's healthcheck.

It seems to work fine in our other deployments though 🤔

[antonsviridov-src]

@jdpleiness How can I replicate this?

I ran docker-compose up from your branch and I can see syntactic worker running and no logs (so assuming healthy)

Spoke too soon – healthcheck was using incorrect port. I made it explicit in d49d846 and fixed in 5c1ad7a

[antonsviridov-src]

@jdpleiness I fixed the build here, is there anything else required?

[jdpleiness]

@jdpleiness I fixed the build here, is there anything else required?

@antonsviridov-src That should be all. I was OOO last week but I'll get back to this today 🙏

[jdpleiness]

@jdpleiness I fixed the build here, is there anything else required?

@antonsviridov-src That should be all. I was OOO last week but I'll get back to this today 🙏

I checked this out locally, everything looks good!