View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000031 | Pgpool-II | Bug | public | 2012-10-23 06:25 | 2013-02-09 11:15 |
| Reporter | jgentsch | Assigned To | t-ishii | ||
| Priority | normal | Severity | crash | Reproducibility | sometimes |
| Status | resolved | Resolution | open | ||
| OS | CentOS | OS Version | 5.8 | ||
| Summary | 0000031: pgpool V3_2_STABLE - segfault in pool_memqcache.c:2529 | ||||
| Description | I'm using shmem and master-slave streaming replication. We're getting segmentation faults occasionally. I'm not sure what it is about the SQL that may make a difference, but I'll include the back trace for the specific child process that died. | ||||
| Tags | No tags attached. | ||||
|
|
|
|
|
I would suspect that it may be the newline characters. |
|
|
I get the same result without the \n and the \". |
|
|
Can you print the value of variables? cache_array cache_array->caches cache_array->num_caches |
|
|
I apologize for the delayed reply. Is this what you're looking for? (gdb) print cache_array $1 = (POOL_QUERY_CACHE_ARRAY *) 0x2ad527c0b010 (gdb) print cache_array->caches $2 = {0x6d64204d4f52460a} (gdb) print cache_array->num_caches $3 = 1162626387 If it helps, here are my cache settings: memory_cache_enabled = on memqcache_method = 'shmem' memqcache_total_size = 1073741824 memqcache_max_num_cache = 1000000 memqcache_expire = 0 memqcache_auto_cache_invalidation = on memqcache_maxcache = 104857600 memqcache_cache_block_size = 104857600 It's a pretty large shmem allocation, but it shouldn't be an issue. Thanks for looking at this. |
|
|
Thanks for the back trace. I found a nasty bug with the module in charge of query cache array handling. The cache arrary is used to keep temporary cache results in a transaction. If there are more than 128 SELECTs in a transaction, the module expands cache_arrary by using realloc. However it does not record the new pointer returned by realloc. So the module keeps on using the old pointer which is absoleted. Attached patch fix the problem. Could you please try it? |
|
|
|
|
|
bug fixed after applying the patch |
|
|
sorry, bug appears again |
|
|
Can you please provide back trace again? |
|
|
|
|
|
Maybe attached file <error_pgpool> help, if not, what i need do to create useful backtrace for you. We use OS Centos 6.3, postgresql-9.2 and pgpool-3.2.1 with your patch |
|
|
error_pgpool doesn't help too much. Can you please create usefull backtrace for me? |
|
|
I can confirm that it fixes my particular crash as of 3.2.2. Thanks for fixing it! |
|
|
You are welcome! |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2012-10-23 06:25 | jgentsch | New Issue | |
| 2012-10-23 06:25 | jgentsch | File Added: pgpool_V3_2_STABLE_shmem_segfault.txt | |
| 2012-10-23 06:28 | jgentsch | Note Added: 0000120 | |
| 2012-10-23 06:54 | jgentsch | Note Added: 0000121 | |
| 2012-10-26 19:09 | t-ishii | Assigned To | => t-ishii |
| 2012-10-26 19:09 | t-ishii | Status | new => assigned |
| 2012-10-27 06:25 | t-ishii | Note Added: 0000126 | |
| 2012-11-03 00:56 | jgentsch | Note Added: 0000137 | |
| 2012-11-09 09:28 | t-ishii | Note Added: 0000142 | |
| 2012-11-09 09:29 | t-ishii | File Added: query_cache.patch | |
| 2012-11-13 22:08 | a.mizurov | Note Added: 0000144 | |
| 2012-11-14 06:12 | a.mizurov | Note Added: 0000145 | |
| 2012-11-14 07:40 | t-ishii | Note Added: 0000146 | |
| 2012-11-15 00:29 | a.mizurov | File Added: error_pgpool.txt | |
| 2012-11-15 00:30 | a.mizurov | Note Added: 0000150 | |
| 2012-11-15 00:34 | a.mizurov | Note Edited: 0000150 | |
| 2012-11-19 16:56 | t-ishii | Note Added: 0000157 | |
| 2012-11-26 16:05 | t-ishii | Status | assigned => feedback |
| 2013-02-09 02:49 | jgentsch | Note Added: 0000224 | |
| 2013-02-09 02:49 | jgentsch | Status | feedback => assigned |
| 2013-02-09 11:15 | t-ishii | Note Added: 0000225 | |
| 2013-02-09 11:15 | t-ishii | Status | assigned => resolved |
