05c437d26a
Support instrumenting the number of database connections
...
Since this is now quite dynamic, it's useful to have a metric for it.
2023-01-01 12:43:06 +00:00
926cb2a5e1
Pull the metrics registry out of the controller
...
This will allow for instrumenting low level database functionality, before
anything starts using the database.
2023-01-01 12:27:34 +00:00
7b69611755
Expose metrics from pg_stats
2022-11-29 11:09:55 +00:00
9f05f5f4f9
Only sometimes attempt manually computing n_distinct values
...
For derivation_inputs.
2022-11-29 10:36:46 +00:00
6ada1cb845
Guard against divide by 0 in update-derivation-outputs-statistics
2022-11-28 13:17:20 +00:00
1a0c5599eb
Do derivation inputs and outputs housekeeping at the end of each job
...
This should help with query performance, as the recursive queries using
derivation_inputs and derivation_outputs are particularly sensitive to the
n_distinct values for these tables.
2022-11-28 11:36:12 +00:00
38b3657013
Use advisory locks to avoid deadlocks during data deletion
...
In the case where multiple data deleting processes end up running at the same
time.
2022-11-28 10:26:46 +00:00
ad93a780d3
Improve the inferior cleanup when computing package derivations
2022-11-24 12:37:49 +00:00
ff6f87a3b9
Skip the derivation linter
...
And remove the chunking of derivation lint warnings.
The derivation linter computes the derivation for each packages supported
systems, but there are two problems with the approach. By doing this for each
package in turn, it forces inefficient uses of caches, since most of the
cached data is only relevant to a single system. More importantly though,
because the work of checking one package is dependent on it's supported
systems, it's unpredictable how much work will happen, and this will tend to
increase as more packages support more systems.
I think especially because of this last point, it's not worth attempting to
keep running the derivation linter at the moment, because it doesn't seem
sustainable. I can't see an way to run it that's futureproof and won't break
at some point in the future when packages in Guix support more systems.
2022-11-24 12:37:49 +00:00
7ae1c97b92
Drop the thread pool idle seconds
...
To hopefully bring down the memory usage from idle connections.
2022-11-24 12:37:45 +00:00
2cf187f10b
Fix calling insert-blocked-builds
2022-11-20 15:44:30 +00:00
e87a8124bf
Render a branch not found page if the branch doesn't exist
2022-11-19 10:07:17 +00:00
91f0fbdeb5
Fix quasiquoting
2022-11-19 10:06:55 +00:00
205f020950
Better guard against exceptions in the build event handlers
2022-11-19 09:46:16 +00:00
ca1e4819b6
Fix closing thread postgresql connections
2022-11-17 16:32:04 +00:00
9fc5821180
Include more information about invalid query parameters
...
In the /compare response.
This should enable qa.guix.gnu.org to detect when the base revision for a
comparison is unknown.
2022-11-17 16:18:28 +00:00
cc61bb5f13
Drop the chunk size when gathering lint warnings
...
To try and bring the peak memory usage down.
2022-11-14 09:26:59 +00:00
ab7df4c6e5
Include blocked_builds information in comparison responses
...
This will make it easier to tell when a scheduled build is yet to start, and
can't start due to a missing dependency.
2022-11-14 09:24:49 +00:00
8294accffe
Remove Build status field from blocking builds page
...
As this is unused.
2022-11-12 11:54:50 +00:00
c46ee47632
Make backfilling blocked_builds a bit smarter
...
And drop the chunk size.
2022-11-12 11:53:14 +00:00
ed114265cd
Handle deleting from blocked_builds when builds are scheduled
...
As scheduling a build might unblock others.
2022-11-12 11:42:26 +00:00
b9305d81a4
View scheduled builds like succeeded builds in terms of blocking
...
This means that an output is viewed to not be blocking if it has a scheduled
build, just as if it has a succeeded build. Also, scheduling builds will
unblock blocked builds.
This is helpful as it means that it reduces noise for blocking builds.
2022-11-12 11:33:37 +00:00
7731c6f340
Tweak backfilling the blocked builds
2022-11-12 10:57:53 +00:00
8e9ab68d14
Use latest_build_status rather than build_status
...
In various places in the blocked-builds module.
2022-11-12 10:57:26 +00:00
a34bf4defc
Spawn specific PostgreSQL connections for the blocked_builds updates
...
So that the queries don't get cancelled by the statement timeout.
2022-11-12 10:46:46 +00:00
fc5f562731
Add index on derivation_outputs id and derivation_id fields
...
This might help with doing recursive queries on the derivations graph.
2022-11-12 10:42:04 +00:00
fb9d99a076
Add extended statistics on package_derivations
...
This helps row count estimates when filtering on system_id and target.
2022-11-12 10:40:43 +00:00
48d8ee885a
Have insert-blocked-builds cache when the partitions exist
...
To make it more efficient.
2022-11-11 11:29:45 +00:00
0f22e3ab40
Rework insert-blocked-builds to make it more efficient
...
This also fixes a typo in the partition name.
2022-11-11 11:29:37 +00:00
4101c6b897
Stop using exception handling when inserting blocked_builds entries
...
As it doesn't work in a transaction.
2022-11-11 10:51:12 +00:00
01e54c36db
Add missing test description
2022-11-11 10:35:09 +00:00
989916b740
Add a blocking builds page
2022-11-11 10:35:09 +00:00
1fb291be40
Add support for incrementally tracking blocked builds
...
This will hopefully provide a less expensive way of finding out if a scheduled
build is probably blocked by other builds failing or being canceled.
By working this out when the build events are recieved, it should be more
feasible to include information about whether builds are likely blocked or not
in various places (e.g. revision comparisons).
2022-11-11 10:35:09 +00:00
95064d39a3
Log heap size when computing system tests
2022-11-06 08:53:04 +01:00
1e2826e095
Add more granular logging for computing system test derivations
2022-11-06 08:47:58 +01:00
d06230fcf4
Close postgresql connections when the thread pool thread is idle
...
I think the idle connections associated with idle threads are still taking up
memory, so especially now that you can configure an arbitrary number of
threads (and thus connections), I think it's good to close them regularly.
2022-10-23 11:28:37 +01:00
aaec813cba
Fix JSON response for unknown revisions
2022-10-17 16:07:46 +01:00
8dd164665a
Expose Guile GC metrics
...
As the guix-data-service process seems to be using excessive amounts of
memory, and this will be useful to track it.
2022-10-09 11:13:28 +01:00
ff77bbea7e
Make it possible to increase the number of thread pool threads
...
And double the default to 16.
2022-10-02 15:08:18 +01:00
ce2e13aa45
Log delays in the thread pool channel
...
As I think with lots of requests, this could become a bottleneck.
2022-10-02 14:57:02 +01:00
640386a84d
Insert guix revision lint warnings in chunks
...
To avoid long running queries.
2022-09-17 08:53:23 +02:00
78a5abee21
Improve chunking when inserting derivation inputs
...
Chunk the values inserted in the query, rather than the derivations involved,
as this is more consistent.
2022-09-17 08:53:23 +02:00
7050ea749f
Reduce some chunk sizes
2022-09-17 00:40:51 +02:00
35cf9ba1bc
Chunk inserting guix revision package derivation entries
2022-09-15 16:25:41 +02:00
8ef896b103
Further reduce some chunk sizes
2022-09-15 16:25:31 +02:00
37e8af60fb
Chunk the data for some queries in insert-missing-data-and-return-all-ids
...
This helps to avoid queries getting logged as slow just because of the amount
of data.
2022-09-15 09:18:19 +01:00
d791c2bd8a
Format some queries generated in insert-missing-data-and-return-all-ids
2022-09-14 15:42:16 +01:00
f41bfcf8b6
Reduce some chunk sizes
...
As these queries are still slow enough to be logged.
2022-09-14 15:42:00 +01:00
189b9a4e86
Implement the JSON response for package reproducibility
2022-09-14 10:52:04 +01:00
ec3dba4894
Implement the JSON response for package substitute availability
2022-09-14 10:37:33 +01:00
