potential deadlock in pkg/domain/sysvar_cache.go:121 domain.(*Domain).rebuildSysVarCache

[YangKeao]

It's produced in CI:

POTENTIAL DEADLOCK:

Previous place where the lock was grabbed

goroutine 132 lock 0xc000ce2658

pkg/domain/sysvar_cache.go:121 domain.(*Domain).rebuildSysVarCache ??? <<<<<

pkg/domain/sysvar_cache.go:51 domain.(*Domain).rebuildSysVarCacheIfNeeded ???

pkg/domain/sysvar_cache.go:62 domain.(*Domain).GetSessionCache ???

pkg/session/session.go:3740 session.(*session).loadCommonGlobalVariablesIfNeeded ???

pkg/session/session.go:2132 session.(*session).ExecuteStmt ???

pkg/session/session.go:1645 session.(*session).ExecuteInternal ???

pkg/domain/domain.go:1630 domain.(*Domain).LoadPrivilegeLoop ???

pkg/session/session.go:3361 session.bootstrapSessionImpl ???

pkg/session/session.go:3282 session.BootstrapSession ???

pkg/session/session.go:3282 session.BootstrapSession ???

tests/realtikvtest/addindextest2/global_sort_test.go:83 addindextest2_test.TestGlobalSortBasic ???


Have been trying to lock it again for more than 20s

goroutine 273 lock 0xc000ce2658

pkg/domain/sysvar_cache.go:121 domain.(*Domain).rebuildSysVarCache ??? <<<<<

pkg/domain/sysvar_cache.go:51 domain.(*Domain).rebuildSysVarCacheIfNeeded ???

pkg/domain/sysvar_cache.go:75 domain.(*Domain).GetGlobalVar ???

pkg/domain/domain.go:1857 domain.(*Domain).LoadBindInfoLoop.(*Domain).globalBindHandleWorkerLoop.func1 ???

pkg/util/wait_group_wrapper.go:96 util.(*WaitGroupEnhancedWrapper).Run.func1 ???

https://do.pingcap.net/jenkins/blue/organizations/jenkins/pingcap%2Ftidb%2Fghpr_check2/detail/ghpr_check2/32519/pipeline/404

[Defined2014]

I think it should be fixed by #40283 and #55407, so close this issue.

[Defined2014]

After going deep into it, I found that I had misjudged it before. However, looking at the code, there is no deadlock in this place. It may be that bootstrap runs too slowly, causing the lock time to be too long. So keep close it.