One of our arbitrary config tests would sometimes fail like this in CI:
```diff
su_nationkey,
cust_nation,
l_year;
- supp_nation | cust_nation | l_year | revenue
----------------------------------------------------------------------
- 9 | C | 2008 | 3.00
-(1 row)
-
+ERROR: cannot connect to localhost:10212 to fetch intermediate results
+CONTEXT: while executing command on localhost:10211
```
When looking at the logs it seems like we were running out of
connections:
```
2022-08-23 14:03:52.856 UTC [28122] FATAL: sorry, too many clients already
2022-08-23 14:03:52.860 UTC [21027] ERROR: cannot connect to localhost:10212 to fetch intermediate results
```
This happened with `CitusThreeWorkersManyShards` config. This test on
purpose tries to push the limits of Citus quite far. And the
`ch_benchmarks_1` test is also run in parallel with a few more ones. So
it's not too weird that it ran out of connections. This doubles the
connection limit in the arbitrary config tests to hopefully not hit this
error again.
Example of failed test: https://app.circleci.com/pipelines/github/citusdata/citus/26365/workflows/7a1b5688-85cc-4bc3-ade5-9bd1d83cd0ed/jobs/747908/parallel-runs/1
(cherry picked from commit 21780b4f65)
This improves debugging of arbitrary configs in two ways:
1. Enable logging of distributed deadlock detection
2. Show output of `psql` commands
(cherry picked from commit a645cb4b94)
In CI our failure_connection_establishment sometimes failed randomly
with the following error:
```diff
-- verify a connection attempt was made to the intercepted node, this would have cause the
-- connection to have been delayed and thus caused a timeout
SELECT * FROM citus.dump_network_traffic() WHERE conn=0;
conn | source | message
------+--------+---------
- 0 | coordinator | [initial message]
-(1 row)
+(0 rows)
SELECT citus.mitmproxy('conn.allow()');
```
Source: https://app.circleci.com/pipelines/github/citusdata/citus/26318/workflows/d3354024-9a67-4b01-9416-5cf79aec6bd8/jobs/745558
The way I fixed this was by removing the dump_network_traffic call. This
might sound simple, but doing this while continuing to let the test
serve its intended purpose required quite some more changes.
This dump_network_traffic call was there because we didn't want to show
warnings in the queries above, because the exact warnings were not
reliable. The main reason this error was not reliable was because we
were using round-robin task assignment. We did the same query twice, so
that it would hit the node with the intercepted connection in one of
those connections. Instead of doing that I'm now using the
"first-replica" policy and do the queries only once. This works, because
the first placements by placementid for each of the used tables are on
the second node, so first-replica will cause the first connection to go
there.
This solved most of the flakyness, but when confirming that the
flakyness was fixed I found some additional errors:
```diff
-- show that INSERT failed
SELECT citus.mitmproxy('conn.allow()');
mitmproxy
-----------
(1 row)
SELECT count(*) FROM single_replicatated WHERE key = 100;
- count
----------------------------------------------------------------------
- 0
-(1 row)
-
+ERROR: could not establish any connections to the node localhost:9060 after 400 ms
RESET client_min_messages;
```
Source: https://app.circleci.com/pipelines/github/citusdata/citus/26321/workflows/fd5f4622-400c-465e-8d82-83f5f55a87ec/jobs/745666
I addressed this with a combination of two things:
1. Only change citus.node_connection_timeout for the queries that we
want to test timeout behaviour for. When those queries are done I
reset the value to the default again.
2. Change our mitm framework to only delay the initial connection packet
instead of all packets. I think sometimes a follow on packet of a previous
connection attempt was causing the next connection attempt to be delayed
even if `conn.allow()` was already called. For our tests we only care about
connection timeouts, so there's no reason to delay any other packets than
the initial connection packet.
Then there was some last flakyness in the exact error that was given:
```diff
-- tests for connectivity checks
SELECT name FROM r1 WHERE id = 2;
WARNING: could not establish any connections to the node localhost:9060 after 900 ms
+WARNING: connection to the remote node localhost:9060 failed with the following error:
name
------
bar
(1 row)
```
Source: https://app.circleci.com/pipelines/github/citusdata/citus/26338/workflows/9610941c-4d01-4f62-84dc-b91abc56c252/jobs/746467
I don't have a good explaination for this slight change in error message, but
given that it is missing the actual error message I expected this to be related
to some small difference in timing: e.g. the server responding to the connection
attempt right after the coordinator determined that the connection timed out.
To solve this last flakyness I increased the connection timeouts and made the
difference between the timeout and the delay a bit bigger. With these tweaks
I wasn't able to reproduce this error on CI anymore.
Finally, I made most of the same changes to failure_failover_to_local_execution,
since it was using the `conn.delay()` mitm method too. The only change that
I left out was the timing increase, since it might not be strictly necessary and
increases time it takes to run the test. If this test ever becomes flaky the first
thing we should try is increase its timeout.
(cherry picked from commit cc7e93a56a)
The failure_single_select test would sometimes fail with an error that's
similar to this:
```diff
-- cancel after first SELECT; txn should fail and nothing should be marked as invalid
SELECT citus.mitmproxy('conn.onQuery(query="^SELECT").cancel(' || pg_backend_pid() || ')');
- mitmproxy
----------------------------------------------------------------------
-
-(1 row)
-
+ERROR: canceling statement due to user request
+CONTEXT: COPY mitmproxy_result, line 1: ""
+SQL statement "COPY mitmproxy_result FROM '/home/circleci/project/src/test/regress/tmp_check/mitmproxy.fifo'"
+PL/pgSQL function citus.mitmproxy(text) line 11 at EXECUTE
BEGIN;
```
This error looked very to the one from #6217 and indeed the cause turned
out to be similar. Because we were canceling all SELECT queries, we
would actually sometimes cancel our mitmproxy SELECT queries itself.
This puts some additional restrictions on the queries that we cancel,
most importantly it should contain the name of the table that we're
selecting from.
I was able to reproduce the original issue locally pretty reliably. With
the changes in this PR it didn't happen again.
In passing this also changes one other failure test that was cancelling
all selects and puts similar additional restrictions on those
cancellations.
Example of failed test in CI: https://app.circleci.com/pipelines/github/citusdata/citus/26305/workflows/4d942b91-f83c-453c-8d9a-ae22d608e756/jobs/745071
(cherry picked from commit 506c16efdf)
The failure_create_distributed_table_non_empty test would sometimes fail
like this:
```diff
-- in the first test, cancel the first connection we sent from the coordinator
SELECT citus.mitmproxy('conn.cancel(' || pg_backend_pid() || ')');
- mitmproxy
----------------------------------------------------------------------
-
-(1 row)
-
+ERROR: canceling statement due to user request
+CONTEXT: COPY mitmproxy_result, line 1: ""
+SQL statement "COPY mitmproxy_result FROM '/home/circleci/project/src/test/regress/tmp_check/mitmproxy.fifo'"
+PL/pgSQL function citus.mitmproxy(text) line 11 at EXECUTE
SELECT create_distributed_table('test_table', 'id');
```
Because the cancel command had no filter it would actually sometimes
cancel the mitmproxy cancel command itself. This PR addresses that by
filtering on CREATE TABLE, which is one of the command that
create_distributed_table will send to the workers.
Example of failing test: https://app.circleci.com/pipelines/github/citusdata/citus/26252/workflows/1b7e5464-cca4-4ec1-99b3-48ddf25c29fa/jobs/742829
(cherry picked from commit e2a24b921e)
Sometimes in CI the columnar_memory test was using slightly more memory
than expected.
```diff
SELECT CASE WHEN 1.0 * TopMemoryContext / :top_post BETWEEN 0.98 AND 1.02 THEN 1 ELSE 1.0 * TopMemoryContext / :top_post END AS top_growth
FROM columnar_test_helpers.columnar_store_memory_stats();
--[ RECORD 1 ]-
-top_growth | 1
+-[ RECORD 1 ]------------------
+top_growth | 1.0206132116232119
-- before this change, max mem usage while executing inserts was 28MB and
```
This PR changes the expectation to be slightly higher, such that this
random increase in memory usage doesn't cause a flaky test.
Failing test: https://app.circleci.com/pipelines/github/citusdata/citus/26256/workflows/c0870f66-3346-4f8d-a1d3-36dfd7c98289/jobs/743028
(cherry picked from commit 4ce17f015b)
Sometimes the columnar_memory test fails in CI with the following error:
```diff
SELECT 1.0 * TopMemoryContext / :top_post BETWEEN 0.98 AND 1.02 AS top_growth_ok
FROM columnar_test_helpers.columnar_store_memory_stats();
-[ RECORD 1 ]-+--
-top_growth_ok | t
+top_growth_ok | f
-- before this change, max mem usage while executing inserts was 28MB and
```
This is almost certainly a harmless failure that simply requires bumping
the margin a little bit. However, it's impossible to say with the
current output. I was unable to reproduce this on-demand on my local
machine or even in CI. So this changes the test to include the actual
value difference in the size of TopMemoryContext when it's outside the
expected range. Then next time it fails we at least have some
information about why.
Example of failing test: https://app.circleci.com/pipelines/github/citusdata/citus/25966/workflows/d472a57b-419a-4f33-b8bc-2e174a98d4d6/jobs/730576
(cherry picked from commit e6a1a86db0)
By running isolation tests in parallel we're just asking for flaky
tasks. The first test might temporarily block one of the commands in the
second test, which we then detect as waiting like this:
```diff
step s2-vacuum-analyze:
VACUUM ANALYZE test_insert_vacuum;
-
+ <waiting ...>
step s1-commit:
COMMIT;
+step s2-vacuum-analyze: <... completed>
```
Debugging flaky tests is also much harder when they are run in parallel.
This PR starts running all our isolation tests sequentially.
The reason for opening this PR was me seeing this failing test:
https://app.circleci.com/pipelines/github/citusdata/citus/26194/workflows/ff57e2cf-8ac4-40fe-bc0c-74a7f8fecb53/jobs/740454
As well as having fixed a similar issue recently in #6122
(cherry picked from commit 85305b2773)
This fixes our most commonly randomly failing failure test. The failing
diff is as follows:
```diff
SELECT citus.mitmproxy('conn.onQuery(query="fetch_intermediate_results").kill()');
mitmproxy
-----------
(1 row)
INSERT INTO target_table SELECT * FROM source_table;
-ERROR: connection to the remote node localhost:xxxxx failed with the following error: connection not open
+ERROR: could not open file "base/pgsql_job_cache/10_0_40/repartitioned_results_20770193413_from_4213590_to_1.data": No such file or directory
+CONTEXT: while executing command on localhost:9060
+while executing command on localhost:57637
SELECT * FROM target_table ORDER BY a;
```
As far as I can tell this is the cause of a race condition: After killing
fetch_intermediate_results on worker 9060, the previously created data
file gets cleaned up. The fetch_intermediate_results call that's sent
to worker 57637 will be cancelled and rolled back soon because of the
failure on the other connection. But if that fetch_intermediate_results
call is able to connect to 9060 before it is cancelled, it won't find
the file it's looking for there anymore. So while it's not the error we
expect, it does indicate that we succeeded.
To avoid this issue instead of killing the fetch_intermediate_results
call directly, we kill the COPY command that it uses to do the fetch.
This results in stable output as can be seen here, where 227 runs of
failure_insert_select_repartition succeeded:
https://app.circleci.com/pipelines/github/citusdata/citus/26168/workflows/9c64a3b6-f46c-4725-9fb4-8f6a2d00a023/jobs/739389
To be clear this changes the test to affects the opposite
fetch_intermediate_results call. This kills the fetch_intermediate_results
call of worker 57637, instead of killing the fetch_intermediate_results call
on worker 9060.
Example of failing test: https://app.circleci.com/pipelines/github/citusdata/citus/26147/workflows/780e95ea-264a-4c9f-ad2e-cf11449a795e/jobs/738467
(cherry picked from commit 8ce12eb51f)
We used to rely on a seperate session to add the coordinator.
However, that might prevent the existing sessions to get
assigned proper gpids, which causes flaky tests.
(cherry picked from commit 961fcff5db)
This removes a flaky test that I introduced in #3868 after I fixed the
issue described in #3622. This test is sometimes fails randomly in CI.
The way it fails indicates that there might be some bug: A connection
breaks after rolling back to a savepoint.
I tried reproducing this issue locally, but I wasn't able to. I don't
understand what causes the failure.
Things that I tried were:
1. Running the test with:
```sql
SET citus.force_max_query_parallelization = true;
```
2. Running the test with:
```sql
SET citus.max_adaptive_executor_pool_size = 1;
```
3. Running the test in parallel with the same tests that it is run in
parallel with in multi_schedule.
None of these allowed me to reproduce the issue locally.
So I think it's time to give on fixing this test and simply remove the
test. The regression that this test protects against seems very unlikely
to reappear, since in #3868 I also added a big comment about the need
for the newly added `UnclaimConnection` call. So, I think the need for
the test is quite small, and removing it will make our CI less flaky.
In case the cause of the bug ever gets found, I tracked the bug in #6189
Example of a failing CI run:
https://app.circleci.com/pipelines/github/citusdata/citus/26098/workflows/f84741d9-13b1-4ae7-9155-c21ed3466951/jobs/736424
For reference the unexpected diff is this (so both warnings and an error):
```diff
INSERT INTO t SELECT i FROM generate_series(1, 100) i;
+WARNING: connection to the remote node localhost:57638 failed with the following error:
+WARNING:
+CONTEXT: while executing command on localhost:57638
+ERROR: connection to the remote node localhost:57638 failed with the following error:
ROLLBACK;
```
This test is also mentioned as the most failing regression test in #5975
(cherry picked from commit d16b458e2a)
This creates consistent test output for isolation tests that involve
`CREATE INDEX CONCURRENTLY`. `CREATE INDEX CONCURRENTLY` is sometimes
temporarily detected as blocking, even though it will complete without any other
queries needing to be run. This change makes sure that we wait until that happens
without running any other queries in the meantime. This way we always get consistent
output. The way we do that is addressed by using an empty step in the same
session as the `CREATE INDEX CONCURRENLTY` command. Doing so forces
the isolation tester to wait until the command is finished and not continue with
steps from other sessions. This is [the recommended approach by Postgres][1].
There's two separate cases which are addressed in slightly different ways:
1. If `CREATE INDEX CONCURRENTLY` is actually blocked on another session: Add an
empty step right after the commit of blocking session.
e.g. `"s2-ddl-create-index-concurrently" "s1-commit" "s2-empty"`
2. If it's not actually blocked on another session: Add [an asterisk marker][2] to make
it look like it's blocked (because sometimes this happens randomly) and right
after that we add an empty step to trigger waiting.
e.g. `"s2-ddl-create-index-concurrently"(*) "s2-empty" "s1-commit"`
In passing this also enables isolation tests that were disabled due to a
bug that has already been fixed for a while.
Fixes#5993
Related to #5910 and #2966
[1]: 5f0adec253/src/test/isolation/README (L197-L204)
[2]: 5f0adec253/src/test/isolation/README (L174-L179)
Co-authored-by: Hanefi Onaldi <Hanefi.Onaldi@microsoft.com>
(cherry picked from commit fd07cc9baf)
This is a continuation of a refactor (with commit sha
2b7cf0c097) that aimed to use Citus helper
UDFs by default in iso tests.
PostgreSQL isolation test infrastructure uses some UDFs to detect
whether concurrent sessions block each other. Citus implements
alternatives to that UDF so that we are able to detect and report
distributed transactions that get blocked on the worker nodes as well.
We needed to explicitly replace PG helper functions with Citus
implementations in each isolation file. Now we replace them by default.
(cherry picked from commit ae58ca5783)
Use Citus helper UDFs by default in iso tests
PostgreSQL isolation test infrastructure uses some UDFs to detect
whether concurrent sessions block each other. Citus implements
alternatives to that UDF so that we are able to detect and report
distributed transactions that get blocked on the worker nodes as well.
We needed to explicitly replace PG helper functions with Citus
implementations in each isolation file. Now we replace them by default.
(cherry picked from commit 2b7cf0c097)
Since #6300/e29db74 changed the C symbol that our bigint overrides of
pg_cancel_backend and pg_terminate_backend called. We needed to do
something to continue to make these functions work after downgrading.
Recreating the old definition with a downgrade scripts is not really
possible, since people are expected to run the downgrade steps when
using the new .so file, which does not contain the old symbols.
So, the easiest way to solve it was also defining the new symbols in our
old Citus versions. Luckily our overrides haven't existed for long, so
these symbol definitions only needed to be backported to 11.0.
* Alter_distributed_table colocateWith:none bug fix for partitioned tables.
* Regression tests added for alter_distributed_table colocateWith:none for partitioned tables
* Update query comparision to be more accurate
(cherry picked from commit 69d2fcf5c0)
DESCRIPTION: Fix reference table lock contention
Dropping and creating reference tables unintentionally blocked on each other due to the use of an ExclusiveLock for both the Drop and conditionally copying existing reference tables to (new) nodes.
The patch does the following:
- Lower lock lever for dropping (reference) tables to `ShareLock` so they don't self conflict
- Treat reference tables and distributed tables equally and acquire the colocation lock when dropping any table that is in a colocation group
- Perform the precondition check for copying reference tables twice, first time with a lower lock that doesn't conflict with anything. Could have been a NoLock, however, in preparation for dropping a colocation group, it is an `AccessShareLock`
During normal operation the first check will always pass and we don't have to escalate that lock. Making it that we won't be blocked on adding and remove reference tables. Only after a node addition the first `create_reference_table` will still need to acquire an `ExclusiveLock` on the colocation group to perform the copy.
There are 3 different ways that a sequence can be interacting
with tables. (1) and (2) are already supported. This commit adds
support for (3).
(1) column DEFAULT nextval('seq'):
The dependency is roughly like below,
and ExpandCitusSupportedTypes() is responsible
for finding the depending sequences.
schema <--- table <--- column <---- default value
^ |
|------------------ sequence <--------|
(2) serial columns: Bigserial/small serial etc:
The dependency is roughly like below,
and ExpandCitusSupportedTypes() is responsible
for finding the depending sequences.
schema <--- table <--- column <---- default value
^ |
| |
sequence <--------|
(3) Sequence OWNED BY table.column: Added support for
this type of resolution in this commit.
The dependency is almost like the following, and
ExpandCitusSupportedTypes() is NOT responsible for finding
the dependency.
schema <--- table <--- column
^
|
sequence
(cherry picked from commit 9ec8e627c1)
For some reason search_path is not always set correctly on the worker
when calling a distributed function, this shows up when calling
`insert_document` in our distributed_triggers test. The underlying
reason is currently unknown and warrants deeper investigation.
Currently this test is one of the main causes for random CI failures. So
this change sets the search_path of each function explicitly, to reduce
these failures. So other devs can be more efficient, while I continue
investigating the root cause of this issue.
Also changes explicit `SET citus.enable_unsafe_triggers = false` to
`RESET citus.enable_unsafe_triggers` in passing.
(cherry picked from commit 6d8c5931d6)
Reported bug #5803 shows that we are currently not sending the IN clause to our planner for columnar. This PR fixes it by checking for ScalarArrayOpExpr in ExtractPushdownClause so that we do not skip it. Also added a test case for this new addition.
It turns out that create_distributed_table
and citus_move/copy_shard_placement does not
work well concurrently.
To fix that, we need to acquire a lock, which
sounds like a good use of colocation lock.
However, the current usage of colocation lock is
limited to higher level UDFs like rebalance_table_shards
etc. Those usage of lock is still useful, but
we cannot acquire the same lock on citus_move_shard_placement
etc. because the coordinator connects to itself to acquire
the lock. Hence, the high level UDF blocks itself.
To fix that, we use one more colocation lock, with the placements
are the main objects to consider.
(cherry picked from commit 12fa3aaf6b)
Before this commit, we required multiple copies of the
same stringInfo if we needed to append/prepend data to
the stringInfo. Now, we optionally get prefix/postfix.
For large string operations, this can save up to %10
memory.
(cherry picked from commit 26fdcb68f0)
Previously, CreateFixPartitionShardIndexNames() created all
the relevant query strings for all the shards, and executed
the large query string. And, in terms of the memory consumption,
this huge command (and its ExprContext generated while running
the command) is the main bottleneck/
With this change, we are reducing the total amount of memory
usage to almost 1/shard_count.
On my local machine, a distributed partitioned table with 120 partitions,
each 32 shards, the total memory consumption reduced from ~3GB
to ~0.1GB. And, the total execution time increased from ~28 seconds
to ~30 seconds. This seems like a good trade-off.
(cherry picked from commit b8008999dc)
DESCRIPTION:
Fix Bug #4949 where Blocking shard moves fails if there is a foreign key between partitioned distributed tables (from child to parent). This is because we try to create constraints before attaching child partitions to parent. This causes constraint failure as parent table will be empty. Fix is to reverse the order i.e. attach partitions before we create constraints.
TESTING:
Added a new test 'shard_move_constraints_blocking' inspired for existing 'shard_move_constraints' where we trigger shard move with 'block_writes' instead of 'force_logical' to add coverage for this scenario.
Jelte Fennema
Önder Kalacı
Hanefi Onaldi
Marco Slot
Gokhan Gulbiz
Marco Slot
Onur Tirtir
Nils Dijk
Ahmet Gedemenli
aykut-bozkurt
Ying Xu
Nitish Upreti