Re: subscription/015_stream sometimes breaks

2024-01-09 Thread Peter Smith
On Tue, Aug 29, 2023 at 10:35 PM Alvaro Herrera wrote: > > On 2023-Aug-29, Amit Kapila wrote: > > > On Mon, Aug 28, 2023 at 5:35 AM Peter Smith wrote: > > > > > > On Fri, Aug 25, 2023 at 8:15 PM Amit Kapila > > > wrote: > > > > > > IMO there are inconsistencies in the second patch that was

Re: subscription/015_stream sometimes breaks

2023-08-29 Thread Alvaro Herrera
On 2023-Aug-29, Amit Kapila wrote: > On Mon, Aug 28, 2023 at 5:35 AM Peter Smith wrote: > > > > On Fri, Aug 25, 2023 at 8:15 PM Amit Kapila wrote: > > > > IMO there are inconsistencies in the second patch that was pushed. > I find your suggestions reasonable. Alvaro, do you have any comments?

Re: subscription/015_stream sometimes breaks

2023-08-28 Thread Amit Kapila
On Mon, Aug 28, 2023 at 5:35 AM Peter Smith wrote: > > On Fri, Aug 25, 2023 at 8:15 PM Amit Kapila wrote: > > IMO there are inconsistencies in the second patch that was pushed. > > 1. In the am_xxx functions, why is there Assert 'in_use' only for the > APPLY / PARALLEL_APPLY workers but not for

Re: subscription/015_stream sometimes breaks

2023-08-27 Thread Peter Smith
On Fri, Aug 25, 2023 at 8:15 PM Amit Kapila wrote: > > On Thu, Aug 24, 2023 at 3:48 PM Amit Kapila wrote: > > > > On Thu, Aug 24, 2023 at 1:20 PM Alvaro Herrera > > wrote: > > > > > > On 2023-Aug-24, Amit Kapila wrote: > > > > > > > On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera > > > >

Re: subscription/015_stream sometimes breaks

2023-08-25 Thread Amit Kapila
On Thu, Aug 24, 2023 at 3:48 PM Amit Kapila wrote: > > On Thu, Aug 24, 2023 at 1:20 PM Alvaro Herrera > wrote: > > > > On 2023-Aug-24, Amit Kapila wrote: > > > > > On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera > > > wrote: > > > > > > Hmm, I think if worker->in_use is false, we shouldn't

Re: subscription/015_stream sometimes breaks

2023-08-24 Thread Amit Kapila
On Fri, Aug 25, 2023 at 9:09 AM Peter Smith wrote: > > On Thu, Aug 24, 2023 at 8:18 PM Amit Kapila wrote: > > > > On Thu, Aug 24, 2023 at 1:20 PM Alvaro Herrera > > wrote: > > > > > > On 2023-Aug-24, Amit Kapila wrote: > > > > > > > On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera > > > >

Re: subscription/015_stream sometimes breaks

2023-08-24 Thread Peter Smith
On Thu, Aug 24, 2023 at 8:18 PM Amit Kapila wrote: > > On Thu, Aug 24, 2023 at 1:20 PM Alvaro Herrera > wrote: > > > > On 2023-Aug-24, Amit Kapila wrote: > > > > > On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera > > > wrote: > > > > > > Hmm, I think if worker->in_use is false, we shouldn't

Re: subscription/015_stream sometimes breaks

2023-08-24 Thread Amit Kapila
On Thu, Aug 24, 2023 at 1:20 PM Alvaro Herrera wrote: > > On 2023-Aug-24, Amit Kapila wrote: > > > On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera > > wrote: > > > > Hmm, I think if worker->in_use is false, we shouldn't consult the rest > > > of the struct at all, so I propose to add the

Re: subscription/015_stream sometimes breaks

2023-08-24 Thread Alvaro Herrera
On 2023-Aug-24, Amit Kapila wrote: > On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera > wrote: > > Hmm, I think if worker->in_use is false, we shouldn't consult the rest > > of the struct at all, so I propose to add the attached 0001 as a minimal > > fix. > > I think that way we may need to add

Re: subscription/015_stream sometimes breaks

2023-08-23 Thread Amit Kapila
On Wed, Aug 23, 2023 at 1:31 PM Alvaro Herrera wrote: > > On 2023-Aug-23, Zhijie Hou (Fujitsu) wrote: > > > [1]-- > > LWLockAcquire(LogicalRepWorkerLock, LW_SHARED); > > > > workers = logicalrep_workers_find(MyLogicalRepWorker->subid, > > true); > >

Re: subscription/015_stream sometimes breaks

2023-08-23 Thread Alvaro Herrera
On 2023-Aug-23, Alvaro Herrera wrote: > And the reason 0002 does not remove the zeroing of ->proc is that the > tests gets stuck when I do that, and the reason for that looks to be > some shoddy coding in WaitForReplicationWorkerAttach, so I propose we > change that too, as in 0003. Hmm,

Re: subscription/015_stream sometimes breaks

2023-08-23 Thread Alvaro Herrera
On 2023-Aug-23, Zhijie Hou (Fujitsu) wrote: > [1]-- > LWLockAcquire(LogicalRepWorkerLock, LW_SHARED); > > workers = logicalrep_workers_find(MyLogicalRepWorker->subid, > true); > foreach(lc, workers) > { >

RE: subscription/015_stream sometimes breaks

2023-08-22 Thread Zhijie Hou (Fujitsu)
> -Original Message- > From: Zhijie Hou (Fujitsu) > Sent: Wednesday, August 23, 2023 10:27 AM > To: Thomas Munro > Cc: Amit Kapila ; pgsql-hackers > > Subject: RE: subscription/015_stream sometimes breaks > > On Wednesday, August 23, 2023 4:55 AM Thomas Mun

Re: subscription/015_stream sometimes breaks

2023-08-22 Thread vignesh C
On Wed, 23 Aug 2023 at 02:25, Thomas Munro wrote: > > On Wed, Aug 23, 2023 at 8:21 AM Thomas Munro wrote: > > I didn't study it closely but it looks like there might be a second > > deadlock, after the one that is expected by the test? Examples from > > the past couple of weeks: > > I should

RE: subscription/015_stream sometimes breaks

2023-08-22 Thread Zhijie Hou (Fujitsu)
On Wednesday, August 23, 2023 4:55 AM Thomas Munro wrote: > > On Wed, Aug 23, 2023 at 8:21 AM Thomas Munro > wrote: > > I didn't study it closely but it looks like there might be a second > > deadlock, after the one that is expected by the test? Examples from > > the past couple of weeks: >

Re: subscription/015_stream sometimes breaks

2023-08-22 Thread Thomas Munro
On Wed, Aug 23, 2023 at 8:21 AM Thomas Munro wrote: > I didn't study it closely but it looks like there might be a second > deadlock, after the one that is expected by the test? Examples from > the past couple of weeks: I should add, it's not correlated with the patches that cfbot is testing,

subscription/015_stream sometimes breaks

2023-08-22 Thread Thomas Munro
Hi, A couple of times a day, cfbot reports an error like this: https://cirrus-ci.com/task/6424286882168832 I didn't study it closely but it looks like there might be a second deadlock, after the one that is expected by the test? Examples from the past couple of weeks: cfbot=> select