Oninit cdi / Archive
Talk to an expert →

Monitor HDR

Posted in 2009

Topics: High Availability & Replication

Mo — — source: Usenet: comp.databases.informix

Informix 11: We run HDR and I see waits on logbf0, logbf1. I think it's something to do with log shipping to secondary node. Is there a way to look at to make sure that log shipping is not the issue why I see some of the high response times on the user sessions. How can I monitor how long log shipping took for a given transaction?

↪ replying to Mo
TheBigPotatoe@Nothere.Co.Uk — — source: Usenet: comp.databases.informix

On Apr 14, 11:40 pm, Mo <mohitanch...@gmail.com> wrote: > Informix 11: > > We run HDR and I see waits on logbf0, logbf1. I think it's something > to do with log shipping to secondary node. Is there a way to look at > to make sure that log shipping is not the issue why I see some of the > high response times on the user sessions. How can I monitor how long > log shipping took for a given transaction? Version? Platform? (O/S and Primary / Secondary RAM / CPU) How many users? (Or more precisely how many open transactions at a checkpoint) What type of logging have you got (unbuffered vs buffered)? What have you got set for LOGBUFF? When the issue occurs check onstat -l and onstat -u onstat -g ath for holders and waiters for LOGBUFF

↪ replying to Mo
jprenaut@yahoo.com — — source: Usenet: comp.databases.informix

On Apr 14, 5:40 pm, Mo <mohitanch...@gmail.com> wrote: > Informix 11: > > We run HDR and I see waits on logbf0, logbf1. I think it's something > to do with log shipping to secondary node. Is there a way to look at > to make sure that log shipping is not the issue why I see some of the > high response times on the user sessions. How can I monitor how long > log shipping took for a given transaction? When you see users waiting on logbf0/logbuf1 do you also see the flusher of the other log buff waiting on the drcb_bqe condition? If that is the case then you have filled all 12 HDR buffers and the primary is currently backed up and unable to send any more data to the secondary. Do you see this all the time, or do you possibly only notice it while checkpoints are happening on your primary? I ask this since I noticed you said you are using version 11. What are your LRU min and max dirty set to for your buffer pools on your primary? Do you have fairly long checkpoint durations that are very small (to 0) block times, and then mostly consist of flush times? Jacques Renaut IBM/Informix ADP

Related threads

Glossary