Thread: BUG #6249: Segmentation fault in VirtualXactLock()
The following bug has been logged online: Bug reference: 6249 Logged by: Fujii Masao Email address: masao.fujii@gmail.com PostgreSQL version: 9.2dev Operating system: Linux hermes 2.6.38-11-generic #50-Ubuntu SMP Mon Sep 12 21:18:14 UTC 2011 i686 i686 i386 GNU/Linux Description: Segmentation fault in VirtualXactLock() Details: When I built Streaming Replication and Hot Standby environment, set max_standby_streaming_delay to 1s and ran the following shell script which creates the conflict between read-only query and recovery, SEGV occurred on the standby. ---------------------- #!/bin/sh # Master's port: 5432 # Standby's port: 9999 psql -c "CREATE TABLE tbl (id int)" -p 5432 psql -c "INSERT INTO tbl VALUES(1)" -p 5432 sleep 1 psql -c "SELECT pg_sleep(60) FROM tbl" -p 9999 & sleep 1 psql -c "DELETE FROM tbl" -p 5432 psql -c "VACUUM tbl" -p 5432 ---------------------- backtrace is: Core was generated by `postgres: startup process recovering 000000'. Program terminated with signal 11, Segmentation fault. #0 0x083114db in VirtualXactLock () (gdb) bt #0 0x083114db in VirtualXactLock () #1 0x083072d1 in ResolveRecoveryConflictWithVirtualXIDs () #2 0x08307355 in ResolveRecoveryConflictWithSnapshot () #3 0x080b8abc in heap_xlog_clean () #4 0x080bc31a in heap2_redo () #5 0x080f02fd in StartupXLOG () #6 0x080f6c70 in StartupProcessMain () #7 0x0811d5f9 in AuxiliaryProcessMain () #8 0x082d17e4 in StartChildProcess () #9 0x082cd151 in PostmasterMain () #10 0x0824d6b4 in main () I was not able to reproduce this issue on 9.1.1. So the bug would exist only on 9.2dev.
On Tue, Oct 11, 2011 at 10:44 AM, Fujii Masao <masao.fujii@gmail.com> wrote: > When I built Streaming Replication and Hot Standby environment, set > max_standby_streaming_delay to 1s and ran the following shell script which > creates the conflict between read-only query and recovery, SEGV occurred on > the standby. In my investigation, the cause is that VirtualXactLock() doesn't handle correctly the case where BackendIdGetProc() returns NULL when the target backend has already ended. Attached patch will fix this issue. Regards, -- Fujii Masao NIPPON TELEGRAPH AND TELEPHONE CORPORATION NTT Open Source Software Center
Attachment
On Tue, Oct 11, 2011 at 3:14 AM, Fujii Masao <masao.fujii@gmail.com> wrote: > On Tue, Oct 11, 2011 at 10:44 AM, Fujii Masao <masao.fujii@gmail.com> wrote: >> When I built Streaming Replication and Hot Standby environment, set >> max_standby_streaming_delay to 1s and ran the following shell script which >> creates the conflict between read-only query and recovery, SEGV occurred on >> the standby. > > In my investigation, the cause is that VirtualXactLock() doesn't > handle correctly > the case where BackendIdGetProc() returns NULL when the target backend has > already ended. Attached patch will fix this issue. Thanks, committed. -- Robert Haas EnterpriseDB: http://www.enterprisedb.com The Enterprise PostgreSQL Company