Open Dany40 opened 1 year ago
Please, check with next snapshot build (4.0.3.2973)
Thank you, I will test it !
Any news ?
Very sorry if I didn't come back before; my report was about an issue we found on our production servers and it is no possible for us to reproduce the stress situation they have; for giving here any new we have to be in front of a new real case in where segments can't be send to the replica side. I will give you any new after a new situation appear. BTW, I didn't see any mention about this thread in the 4.0.3 release notes. Is this fix included?
The bug is nor confirmed, nor marked as fixed. Thus it is not included into RN. Anyway, the supposed fix was committed into v4 branch and is present in 4.0.3 release.
No news ?
@Dany40 - any feedback ?
Hello;
Firebird 4 on Windows.
In our case, our databases are doing async replication and we defined a command line for running a ftp client to make Firebird to send each segment to a remote machine. Our ftp client returns non cero value if it can't send a segment and Firebird will try again and again, up to it gets a cero value, and this is working very well. The big problem is that in the while, Firebird starts adding garbage in memory. For example, our machine is always at 16-19% of memory usage; if the ftp client cannot send for one our, the machine memory will go to 35-40% of memory usage. Let me add that we have 90 databases connected, and 50 to 120 users connected all the time, doing transactions.
Let me know if I can add more information.