-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
roachtest: schemachange/mixed/tpcc failed #40935
Comments
This is scary... is this a new failure mode? I don't think that FK check should ever fail. cc @lucy-zhang |
Okay, this failure is actually coming from the
Still, this |
SHA: https://github.com/cockroachdb/cockroach/commits/d6a7e59e653596b8baca946b6be714956a0e4c2c Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1499672&tab=artifacts#/schemachange/mixed/tpcc
|
SHA: https://github.com/cockroachdb/cockroach/commits/a92c7d01d3076eabafbd536d8a344511ec9081c6 Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1500206&tab=artifacts#/schemachange/mixed/tpcc
|
SHA: https://github.com/cockroachdb/cockroach/commits/6b14c0aa3ed1b4ba6d5f937e9352c5383afe1c37 Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1501050&tab=artifacts#/schemachange/mixed/tpcc
|
Unrelatedly to the failure on master, this is also failing on 19.1 because my attempt to fix the test for 19.1 (#40912) caused more problems: since the table created using |
SHA: https://github.com/cockroachdb/cockroach/commits/6b14c0aa3ed1b4ba6d5f937e9352c5383afe1c37 Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1502387&tab=artifacts#/schemachange/mixed/tpcc
|
SHA: https://github.com/cockroachdb/cockroach/commits/4b6ab256d09e9a03e68156a3504e08f249cd0af1 Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1505090&tab=artifacts#/schemachange/mixed/tpcc
|
41079: roachtest: fix schemachange/mixed/tpcc for 19.1 r=lucy-zhang a=lucy-zhang `schemachange/mixed/tpcc` uses `CREATE TABLE AS` in 19.2. This PR will have the test correctly create a similar table in 19.1 without using `CREATE TABLE AS`. See #40935 (comment). Release justification: Fixes a test. Release note: None Co-authored-by: Lucy Zhang <[email protected]>
The FK constraint that's being violated is
and the rows reported as invalid in
which doesn't make sense, since rows in I'm stressing this roachtest right now, but haven't been able to reproduce these failures yet. |
Got a successful reproduction with a significantly shorter version of this test:
(this diff is against 073999b)
|
The patch I posted has produced this failure in 5/10 runs. I checked two of the clusters, and the relevant rows in |
SHA: https://github.com/cockroachdb/cockroach/commits/77f26d185efb436aaac88243de19a27caa5da9b6 Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1509340&tab=artifacts#/schemachange/mixed/tpcc
|
SHA: https://github.com/cockroachdb/cockroach/commits/fb0bd2f43fb87b26752615bbdbc759502c2c8b0b Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1509988&tab=artifacts#/schemachange/mixed/tpcc
|
SHA: https://github.com/cockroachdb/cockroach/commits/cf5c2bd2372e633d2f63e08e5bffca7c2a7ec59f Parameters: To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1510142&tab=artifacts#/schemachange/mixed/tpcc
|
This also failed for the beta release with the fk violation. What's the current thinking on it? It's bad? |
Yeah, the failures on 19.1 are due to a trivial problem with the test, which is now fixed. The real problem is the foreign key violations reported by |
I've been able to reproduce this failure by running only the I'm currently trying out replacing the |
(roachtest).schemachange/mixed/tpcc failed on master@14094e3d5ea5f548dfada7bdb6e0e1158f53e168:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@5f9a71adb995837bcff27b9456188018434be4b8:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@c473f40078994551cebcbe00fdbf1fa388957658:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@69dc87d68addedf2fabfb2b14c098cfb35b5f3d0:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@f5585e933a097b53242d8d5800127b821a9a4d41:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@d8bdc938177ea4b5226b95ee6cc51af0951054fc:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@d571624ae4f833c49c717728a74cc7be78a791f0:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@2032dafccfa311c7538960e974953cb9dc1d4e50:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
(roachtest).schemachange/mixed/tpcc failed on master@7b0f60fe2034ba8677242dbcdad86d3e5587c0d4:
More
Artifacts: /schemachange/mixed/tpcc
See this test on roachdash |
We also saw this in #47471. Let's track it there. |
47492: kv: respect exhausted key limit during ranged intent resolution r=nvanbenschoten a=nvanbenschoten Fixes #47471. Fixes #40935. This commit fixes a long-standing bug where ranged intent resolution would not respect the MaxSpanRequestKeys set on a batch once the limit had already been exhausted by other requests in the same batch. Instead of treating the limit as exhausted, ranged intent resolution would consider the limit nonexistent (unbounded). This bug was triggering an assertion in DistSender. We became more likely to hit this issue in v20.1 because we started performing ranged intent resolution more often due to implicit SELECT FOR UPDATE. This commit fixes the bug in two ways: 1. it addresses the root cause, updating MVCCResolveWriteIntentRangeUsingIter to properly respect the limit placed on the request when it is exhauted. 2. it disables the assertion in DistSender when it detects that we are hitting this bug. This ensures that we don't hit the assertion in mixed version clusters (see #40935). By the time we're in DistSender, the damage is already done and has already potentially resulted in a large Raft entry. Maintaining the assertion doesn't do us any good. Release notes (bug fix): a bug that could could trigger an assertion with the text "received X results, limit was Y" has been fixed. The underlying bug was only performance related and could not cause user-visible correctness violations. Release justification: fixes a medium-priority bug in existing functionality. The bug could result in an assertion failure and a node crashing. Even though this was an old bug (present in many releases before v20.1), it became a lot easier to hit in v20.1 because we started performing ranged intent resolution more often due to implicit SELECT FOR UPDATE. Co-authored-by: Nathan VanBenschoten <[email protected]>
Fixes cockroachdb#47471. Fixes cockroachdb#40935. This commit fixes a long-standing bug where ranged intent resolution would not respect the MaxSpanRequestKeys set on a batch once the limit had already been exhausted by other requests in the same batch. Instead of treating the limit as exhausted, ranged intent resolution would consider the limit nonexistent (unbounded). This bug was triggering an assertion in DistSender. We became more likely to hit this issue in v20.1 because we started performing ranged intent resolution more often due to implicit SELECT FOR UPDATE. This commit fixes the bug in two ways: 1. it addresses the root cause, updating MVCCResolveWriteIntentRangeUsingIter to properly respect the limit placed on the request when it is exhauted. 2. it disables the assertion in DistSender when it detects that we are hitting this bug. This ensures that we don't hit the assertion in mixed version clusters (see cockroachdb#40935). By the time we're in DistSender, the damage is already done and has already potentially resulted in a large Raft entry. Maintaining the assertion doesn't do us any good. Release notes (bug fix): a bug that could could trigger an assertion with the text "received X results, limit was Y" has been fixed. The underlying bug was only performance related and could not cause user-visible correctness violations. Release justification: fixes a medium-priority bug in existing functionality. The bug could result in an assertion failure and a node crashing. Even though this was an old bug (present in many releases before v20.1), it became a lot easier to hit in v20.1 because we started performing ranged intent resolution more often due to implicit SELECT FOR UPDATE.
SHA: https://github.com/cockroachdb/cockroach/commits/073999b81ddfed3bbc8409d534912fea12b6d500
Parameters:
To repro, try:
Failed test: https://teamcity.cockroachdb.com/viewLog.html?buildId=1498740&tab=artifacts#/schemachange/mixed/tpcc
The text was updated successfully, but these errors were encountered: