Jehan-Guillaume (ioguix) de Rorthais

Hi, I created the same plan in verbose mode, to include JIT infos in workers: https://explain.dalibo.com/plan/9hO JIT info related to the bad reported seqscan timing are reported there: https://explain.dalibo.com/plan/9hO#plan/node/5 Most...

Hello @YanChii, This is something I had in mind for a while as well. However, setting attributes to the CIB during transition breaks it. That means we will loose the...

@YanChii, About your auto-failback project, the best way to detect a failed master not being able to catchup with the new master it to compare its TL+LSN with TL history....

> It is true that I'm seeing in logs: > ``` > Jan 23 10:30:45 cluster6 crmd[5566]: notice: Transition aborted by cib-bootstrap-options-postgres-6366165775826891940-highest-timeline doing modify postgres-6366165775826891940-highest-timeline=31: Non-status change > ``` >...

Hi, I started a new branch on my repo to work on the TL check during failover for a safer election code. You can find a diff of my current...

Hi @YanChii , Following our discussion on IRC, I started a branch in my repository to allow more than one PAF resource in the same cluster. My limited tests so...

Hi, As reported in #194, this is not the (only?) root case of the issue. I'm waiting for some more investigation before we can settle an appropriate fix. Thanks for...

We discussed the `--get-value` issue on pacemaker IRC chan. The argument is deprecated, but it seems this behavior is a regression as well. Upstream is taking care of both bugs...

Hi, Master scores are delivered by the current master of your cluster based on its local data from `pg_stat_replication`. The scores are sorted by the highest `write_lsn` or (`write_location` for...

In fact, there is a lot a asynchronous piece in this puzzle: * the standbies feedbacks are not in simultaneous * the standbies sync of WAL on disk are not...