At Wed, 13 May 2020 15:10:30 -0400, Robert Haas <robertmh...@gmail.com> wrote in > On Tue, May 12, 2020 at 10:26 PM Amit Kapila <amit.kapil...@gmail.com> wrote: > > I was trying to say that tablespace listing will happen under > > PROGRESS_BASEBACKUP_PHASE_WAIT_CHECKPOINT phase which could be a > > problem if it is a costly operation but as you said it is pretty cheap > > so I think we don't need to bother about that. > > > > Apart from the above point which I think we don't need to bother, both > > your patches look good to me. > > OK, good. Let's see if anyone else feels differently about this issue > or wants to raise anything else. If not, I'll plan to commit these > patches after we branch. Thanks for the review.
Table space listing needs only one or few 512k pages, which should be on OS file cache, which cannot take long time unless the system is facing a severe trouble. (I believe that is the same on Windows.) I'm fine that WAIT_CHECKPOINT contains the time to enumerate tablespace directories. 0001 looks good to me. The progress information gets About 0002, + bool sendtblspclinks = true; The boolean seems to me useless since it is always the inverse of opt->sendtblspcmapfile when it is used. Everything looks fine to me except the above. regards. -- Kyotaro Horiguchi NTT Open Source Software Center
