[Progress Communities] [Progress OpenEdge ABL] Forum Post: Buffer locked (or not locked?) on target database

[George Potemkin]

Target database is used by processes (appsevers) to provide a service. At some point of time we got a lot of blocked clients (‘Status: Blocked Clients’). Almost all connected processes were waited for the same dbkey – the root block of the most active index. The clients stayed blocked during 8 minutes. BUT ‘Status: Buffer Locks’ screen in promon checked many times during the same interval did not report any buffer locks. 06/25/19 Status: Blocked Clients by user number for all tenants Sample Date Time Usr:Ten Name Domain Type Wait Wait Info Trans id Num Txns BI RRead BI RWries Login time Schema Timestamp 5671 06/25/19 19:52:59 5595 root -4 REMC/APSV BKSH 27375479:8 0 0 0 0 06/25/19 19:41 1526776349 0 5671 06/25/19 19:52:59 5597 root -4 REMC/APSV BKSH 27375479:8 0 0 0 0 06/25/19 19:36 1526776349 0 ... 5712 06/25/19 19:59:55 5631 root -4 REMC/APSV BKSH 27375479:8 0 0 0 0 06/25/19 19:29 1526776349 0 5712 06/25/19 19:59:55 5632 root -4 REMC/APSV BKSH 27375479:8 0 0 0 0 06/25/19 19:43 1526776349 0 Approximately at the same time replication agent started new transaction and there was no BI write activity (BI RWrites did not incremented) during the same interval: 06/25/19 Status: Active Transactions by user number for all tenants Sample Date Time Usr:Ten Name Domain Type Login time Tx start time Trans id BI RReads BI RWrites Trans State 5669 06/25/19 19:52:39 513 root 0 RPLA 06/16/19 04:26 - 1124543754 0 343759567 None FWD 5670 06/25/19 19:52:49 513 root 0 RPLA 06/16/19 04:26 - 1124543757 0 343759581 None FWD 5671 06/25/19 19:52:59 513 root 0 RPLA 06/16/19 04:26 06/25/19 19:52 1124543758 0 343759755 Active FWD ... 5712 06/25/19 19:59:55 513 root 0 RPLA 06/16/19 04:26 06/25/19 19:52 1124543758 0 343759755 Active FWD 5713 06/25/19 20:00:05 513 root 0 RPLA 06/16/19 04:26 06/25/19 19:52 1124543758 0 343759881 Active FWD The questions: 1. When a process is waiting for buffer lock does it mean that the buffer is indeed locked rather than other database resources (for example, the lathes) are locked? 2. Can ‘Status: Buffer Locks’ screen miss some buffer locks? It scans Usrctrl Table rather than buffer pool (Buffer Headers) scanned, for example, by ‘Status: Buffer Lock Queue’. Unfortunately, promon is broken since 11.7 (OCTA-13662) and we are forced to limit the volume of data gathered through promon - we are not yet totally blind but we do have the “vision problems”. For example, I don’t know if any latches were locked during this interval. The customer complained about significant degradation of the service response time.

Continue reading...