[Enhancement] Report load profile if brpc reaches timeout #55494
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
4bbc693 to
f7b13c9
Compare
d5dab22 to
cbfba93
Compare
wyb
reviewed
Feb 18, 2025
| const starrocks::PLoadDiagnoseRequest* request, | ||
| starrocks::PLoadDiagnoseResult* response, | ||
| google::protobuf::Closure* done) { | ||
| VLOG_RPC << "load diagnose, id=" << print_id(request->id()) << ", txn_id: " << request->txn_id(); |
There was a problem hiding this comment.
Suggested change
| VLOG_RPC << "load diagnose, id=" << print_id(request->id()) << ", txn_id: " << request->txn_id(); | |
| VLOG_RPC << "load diagnose, id=" << print_id(request->id()) << ", txn_id=" << request->txn_id(); |
| @@ -2224,7 +2225,7 @@ public void handleDMLStmtWithProfile(ExecPlan execPlan, DmlStmt stmt) throws Exc | |||
| throw t; | |||
| } finally { | |||
| boolean isAsync = false; | |||
| if (context.isProfileEnabled()) { | |||
| if (context.isProfileEnabled() || LoadErrorUtils.enableProfileAfterError(coord)) { | |||
There was a problem hiding this comment.
does stream load need to check LoadErrorUtils.enableProfileAfterError?
There was a problem hiding this comment.
currently stream load always reports profile after failure no matter what's the reason, so no need to check LoadErrorUtils.enableProfileAfterError for stream load
| // The timeout of the diagnosis rpc sent from OlapTableSink to LoadChannel | ||
| CONF_mInt32(load_diagnose_send_rpc_timeout_ms, "2000"); | ||
| // Used in load fail point. The brpc timeout used to simulate brpc exception "[E1008]Reached timeout" | ||
| CONF_mInt32(load_fp_brpc_timeout_ms, "-1"); |
There was a problem hiding this comment.
maybe we support timeout in admin enable failpoint script later?
There was a problem hiding this comment.
as discussed offline, maybe do it later
Signed-off-by: PengFei Li <lpengfei2016@gmail.com>
Signed-off-by: PengFei Li <lpengfei2016@gmail.com>
cbfba93 to
a68e6fb
Compare
|
[Java-Extensions Incremental Coverage Report]✅ pass : 0 / 0 (0%) |
[FE Incremental Coverage Report]✅ pass : 15 / 17 (88.24%) file detail
|
[BE Incremental Coverage Report]✅ pass : 104 / 119 (87.39%) file detail
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Why I'm doing:
A common issue encountered during import is brpc timeout, such as
[E1008]Reached timeout 150000ms@10.128.8.78:8060, which is often caused by prolonged durations at certain stages on the storage side. Currently, there is a lack of information to analyze the time consumption when such problems occur. Therefore, it is desirable to provide an observability mechanism that, upon detecting a "brpc reached timeout" exception, can proactively generate a profile containing some time consumption statistics from the storage side, thereby aiding in pinpointing the cause.What I'm doing:
The brpc is sent from the Coordinator BE (OlapTableSink) to the Executor BE (LoadChannel) on the storage side. A mechanism is to be added within OlapTableSink that, upon detecting a brpc timeout, actively sends a diagnose rpc to the LoadChannel, requesting the current profile from the LoadChannel, which is then reported to the FE.
Fixes #issue
What type of PR is this:
Does this PR entail a change in behavior?
If yes, please specify the type of change:
Checklist:
Bugfix cherry-pick branch check: