[pgpool-hackers: 2341] memory leak in "On Memory Query Cache"?

Huong Dangminh huo-dangminh at ys.jp.nec.com
Wed May 17 15:02:10 JST 2017


Hi,

We have tested using of memory when memory_cache_enabled=on and found that,
RES(from top command) of pgpool process is continuously increasing.

We will try to help to fix if this is a bug,
but first we want to confirm is this a specification or not?
# I mean, is there a threshold for that increasing memory?

--- Addition information ---
It occur in both 'shmem' and 'memcached' mode.
We could not reproduce this problem when run pgbench with -S (SELECT only) option.
So, might be memory leak somewhere in discard or create cache processing? 

--- reproduce information ---
version
---
pgpool-II 3.4.11
PostgreSQL 9.4.4
---

pgpool.conf
---
...
master_slave_mode = off
master_slave_mode = off
memory_cache_enabled = on
memqcache_method = 'shmem' # it is also reproduced in 'memcached' mode
memqcache_total_size = 15728640 # about 15MB
...
---

test result (top result)
---
start(pgbench -c 1 -T 10000):

 PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
14499 root      20   0  117m 9988 1376 R 25.0  1.0   0:08.91 pgpool: testuser testdb  UPDATE
17135 postgres  20   0  122m 5148 3432 S 10.0  0.5   0:03.54 postgres: testuser testdb pgpoolhost(35249) idle in transaction

after about 10 minutes:

14499 root      20   0  338m 230m 1268 R  6.9 23.1  11:45.71 pgpool: testuser testdb  SELECT
17135 postgres  20   0  122m 5176 3460 D  3.3  0.5   4:53.74 postgres: testuser testdb pgpoolhost(35249) COMMIT

Only RES of pgpool (but not SHR) had increased about 200MB and was continuously increasing until connection from pgbench disconnected.
---

--- 
Thanks and best regards, 
Dang Minh Huong 
NEC Solution Innovators, Ltd. 
http://www.nec-solutioninnovators.co.jp/en/



More information about the pgpool-hackers mailing list