View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0000378 | Pgpool-II | Bug | public | 2018-02-05 23:57 | 2018-05-07 15:29 |
| Reporter | tuomas | Assigned To | t-ishii | ||
| Priority | normal | Severity | major | Reproducibility | have not tried |
| Status | closed | Resolution | open | ||
| Platform | Linux | OS | Ubuntu | OS Version | 16.04.3 |
| Product Version | 3.6.7 | ||||
| Summary | 0000378: Pgpool fails all connections when secondary goes down | ||||
| Description | Configuration: - 2 backends. backend 0 is secondary, backend 1 is primary - failovers are disallowed - streaming replication mode - loadbalancing is enabled When backend 0 (secondary) is restarted/stopped, all connections to pgpool fail with error: ERROR: unable to read message kind DETAIL: kind does not match between master(45) slot[1] (52) In pgpool logs: 2018-02-05 14:20:59: pid 31248: DETAIL: postmaster on DB node 0 was shutdown by administrative command 2018-02-05 14:20:59: pid 29420: LOG: pool_send_and_wait: Error or notice message from backend: : DB node id: 0 backend pid: 2707 statement: " DISCARD ALL" message: "terminating co nnection due to administrator command" 2018-02-05 14:20:59: pid 29420: WARNING: packet kind of backend 1 ['C'] does not match with master/majority nodes packet kind ['E'] 2018-02-05 14:20:59: pid 29420: FATAL: failed to read kind from backend 2018-02-05 14:20:59: pid 29420: DETAIL: kind mismatch among backends. Possible last query was: " DISCARD ALL" kind details are: 0[E: terminating connection due to administrator co mmand] 1[C] 2018-02-05 14:20:59: pid 29420: HINT: check data consistency among db nodes 2018-02-05 14:20:59: pid 31365: LOG: reading and processing packets 2018-02-05 14:20:59: pid 31365: DETAIL: postmaster on DB node 0 was shutdown by administrative command 2018-02-05 14:20:59: pid 31365: LOG: degenerate backend request for node_id: 0 from pid [31365] is canceled because failover is disallowed on the node2018-02-05 14:20:59: pid 29370: LOG: pool_read_kind: error message from master backend:the database system is shutting down 2018-02-05 14:20:59: pid 29370: ERROR: unable to read message kind 2018-02-05 14:20:59: pid 29370: DETAIL: kind does not match between master(45) slot[1] (52) ... | ||||
| Steps To Reproduce | Possibly by setting up 2 postgresql servers in streaming replication and then configuring pgpool so that backend 0 is secondary and backend 1 is primary and failovers are disabled and then shutting down the secondary postgresql. I can try to set up some test environment if needed, but the error messages might already give enough information. | ||||
| Tags | No tags attached. | ||||
|
|
I was not able to reproduce exactly the same issue you had. Instead, I got segfault of child process. Both are not good of course :--< Will look into this.... |
|
|
I forgot to mention that the backends are postgresql 9.5, it might make a difference |
|
|
Regarding "failovers are disallowed". What do you set to pgpool.conf exactly? |
|
|
No response over 1 month. I'm going to close this issue. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2018-02-05 23:57 | tuomas | New Issue | |
| 2018-02-06 11:06 | t-ishii | Assigned To | => t-ishii |
| 2018-02-06 11:06 | t-ishii | Status | new => assigned |
| 2018-02-06 11:07 | t-ishii | Note Added: 0001917 | |
| 2018-02-08 18:24 | tuomas | Note Added: 0001921 | |
| 2018-02-16 13:39 | t-ishii | Note Added: 0001935 | |
| 2018-02-16 16:52 | t-ishii | Status | assigned => feedback |
| 2018-04-06 10:50 | t-ishii | Note Added: 0001986 | |
| 2018-05-07 15:29 | t-ishii | Status | feedback => closed |
