Roland Olbricht
Roland Olbricht
As there are organizational issues with the next server, I have configured now `dev` such that there is a clone at least three times a week.
Implemented in ec0a66834588c93273f5cc92c53ebeb1b689c2cd Not yet released.
The change in the block size is something that one needs to actively trigger. So I would be surprised if this comes into play here. With a command like for...
Thank you for the data on gorwen. Well, it is expected that 128k blocks on HDD will strain the disk. The size of 512k have been the found optimum a...
The intended result of this issue should be a general patch version that helps gorwen, instructions how to best configure a magnetic disk system or both. The fist step for...
It turned out that with an extra syscall `posix_madvise` the OS can be directed to perform the same disk caching strategy as before. Therefore, I highly recommend to install 0.7.59.1...
Thank you for the quick update. I agree that this does not look as if it had solved the right problem on gorwen. The dent on the disk utilization on...
Could you please extract from the transactions.log as far back as possible the numbers for the read blocks per update? I suggest the command cat $DB_DIR/transactions.log | grep -E 'write_commit.....
Thank you for the log files. So from the app's perspective, everything looks fine. During the nights it is able to read up 1000 blocks per second, but during the...
> I need to double check my log files, but iirc, database lag starts increasing over night on the dev instance. My assumption here is that the clone process is...