View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000223 | Pgpool-II | Bug | public | 2016-07-26 22:03 | 2016-12-20 14:03 |
| Reporter | rmwpl | Assigned To | t-ishii | ||
| Priority | high | Severity | major | Reproducibility | sometimes |
| Status | closed | Resolution | open | ||
| Platform | x86_64 | OS | GNU/Linux | OS Version | CentOS 6.6 |
| Summary | 0000223: PgPool performs load balancing even when it's disabled | ||||
| Description | Hi, I'm running PgPool with: load_balance_mode = off master_slave_mode = on master_slave_sub_mode = 'stream' I have 4 nodes. Node ID=1 is the primary. Most of the time everything works fine - new connections are routed to the primary DB node. But sometimes I'm getting an error like this: "ERROR - DBIx::Class::Storage::DBI::_dbh_execute(): DBI Exception: DBD::Pg::st execute failed: FATAL: unable to read data from DB node 0 DETAIL: EOF encountered with backend FATAL: unable to read data from DB node 0 DETAIL: EOF encountered with backend [for Statement "INSERT INTO public...." Basically, although load balancing is specifically disabled, PgPool routes some INSERT queries to a standby DB node. We've been doing a lot of INSERTs and it was all fine. But few weeks ago we've disabled connection_cache. It's the only change made to PgPool's configuration lately. | ||||
| Steps To Reproduce | Set master_slave_mode to on/stream and load_balance_mode to off. Have 4 nodes, do a lot of INSERTs. Probably also disable connection_cache. | ||||
| Tags | connection cache, load balancing, master slave, streaming replication | ||||
|
|
Which version of pgpool-II? |
|
|
pgpool-II-95-3.4.3-4.rhel6.x86_64 |
|
|
It's not the rpm file name I know of. Anyway, I guess it's too old version. Please upgrade to the latest version (hint: 3.4.7). |
|
|
# pgpool --version pgpool-II version 3.4.3 (tataraboshi) |
|
|
Do you mean that's a known bug that has been fixed in the latest version? |
|
|
Oh, I see a potential bug that might have caused the problem for me. Do you gather it's the same case as in the "Deal with the case when the primary is not node 0 in streaming replication mode. (Tatsuo Ishii)" changelog item? |
|
|
Yes, that's my thought. |
|
|
Thanks, that's helpful. I will upgrade then and report back if the error persists. |
|
|
Tried upgrading to 3.5.3, but it doesn't work properly. I brings the delegate ip up and after a seconds it brings it down and shuts down. |
|
|
Do you know how to reproduce the problem? Without knowing it, it's pretty hard to debug... |
|
|
No action from the reporter over 1 month. Issue closed. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2016-07-26 22:03 | rmwpl | New Issue | |
| 2016-07-26 22:03 | rmwpl | Tag Attached: connection cache | |
| 2016-07-26 22:03 | rmwpl | Tag Attached: load balancing | |
| 2016-07-26 22:03 | rmwpl | Tag Attached: master slave | |
| 2016-07-26 22:03 | rmwpl | Tag Attached: streaming replication | |
| 2016-07-26 23:33 | t-ishii | Assigned To | => t-ishii |
| 2016-07-26 23:33 | t-ishii | Status | new => assigned |
| 2016-07-26 23:34 | t-ishii | Note Added: 0000933 | |
| 2016-07-26 23:51 | rmwpl | Note Added: 0000934 | |
| 2016-07-27 00:11 | t-ishii | Note Added: 0000935 | |
| 2016-07-27 00:17 | rmwpl | Note Added: 0000936 | |
| 2016-07-27 00:18 | rmwpl | Note Added: 0000937 | |
| 2016-07-27 00:22 | rmwpl | Note Added: 0000938 | |
| 2016-07-27 00:26 | t-ishii | Note Added: 0000939 | |
| 2016-07-27 00:29 | rmwpl | Note Added: 0000940 | |
| 2016-07-27 07:44 | t-ishii | Status | assigned => feedback |
| 2016-08-25 20:43 | rmwpl | Note Added: 0001023 | |
| 2016-08-25 20:43 | rmwpl | Status | feedback => assigned |
| 2016-09-23 15:28 | t-ishii | Note Added: 0001082 | |
| 2016-09-23 15:28 | t-ishii | Status | assigned => feedback |
| 2016-12-20 14:03 | t-ishii | Note Added: 0001240 | |
| 2016-12-20 14:03 | t-ishii | Status | feedback => closed |
