301 reputation
117
bio website syneticon.net
location Germany
age
visits member for 2 years, 9 months
seen Oct 2 at 21:11

I am working for a company which is doing infrastructure operations and IT consulting jobs in Germany, Europe.

My sphere of activities includes virtualization topics, IP networking, storage, server operations, monitoring, backups and network security. I was awarded the Microsoft MVP title for Windows Networking for several years in the past, although my public activity on Microsoft-specific topics has declined seriously due to lack of time so it is not likely to happen again that soon.

If you need to get in touch, just mail the company: office@syneticon.net

If you are wondering about the avatar bunny, Ubisoft supplies a lot more to look at and wonder about.


Oct
2
comment Sudden periods of performance degradation
The older release might have bugs and performance-related issues which have been fixed in the newer release and thus were not visible prior to the role-swap. As replication is asymmetric and slave-initiated, an older slave is not going to affect the master performance-wise.
Oct
2
comment Sudden periods of performance degradation
Possibly related: dba.stackexchange.com/questions/75439/…. You can take a look at what has been done for diagnosis so you could use this in your own setup even if the premise of full-text search indexes does not apply to your setup.
Sep
24
awarded  Autobiographer
Sep
6
comment MySQL - HOWTO speed up slave replication on large database / Quickly spawn slave?
The increasing replication lag might be due to replication errors. You should check your mysql server's error.log files and double-check that you have your IDs (server-id and, if applicable, server-uuid) are indeed unique throughout your replication topology.
Sep
5
revised MySQL instance stalling “doing SYNC index”
added 49 characters in body
Sep
5
revised MySQL instance stalling “doing SYNC index”
added 689 characters in body
Sep
5
revised MySQL instance stalling “doing SYNC index”
edited body
Sep
5
revised MySQL instance stalling “doing SYNC index”
added 588 characters in body
Sep
5
comment MySQL instance stalling “doing SYNC index”
ping - I suppose I have found the trigger being an update to the FTS index. Now I just need to figure out why it happens.
Sep
5
revised MySQL instance stalling “doing SYNC index”
added 879 characters in body; edited title
Sep
4
awarded  Nice Question
Sep
4
revised MySQL instance stalling “doing SYNC index”
added 100 characters in body
Sep
4
comment MySQL instance stalling “doing SYNC index”
Ronaldo, thanks, I already tried running O_DIRECT, just forgot to mention it in the respective section. @James no, writes are not stalling. MySQL itself is doing a number of writes at times of stalls (to the logs apparently), but it is waiting for something.
Sep
4
revised MySQL instance stalling “doing SYNC index”
added 1331 characters in body; edited title
Sep
3
revised MySQL instance stalling “doing SYNC index”
added 1841 characters in body
Sep
3
comment MySQL instance stalling “doing SYNC index”
I am through changing most of the parameters you've suggested except for innodb_purge_threads (which needs a restart). It did not do much for the issue. And I am led to believe that the InnoDB engine is not the problem here as MyISAM table inserts are stalling too.
Sep
3
revised MySQL instance stalling “doing SYNC index”
added 680 characters in body; edited tags
Sep
3
comment Configuration to stop mysql sleep
The "connection reset" SocketException is telling us that your database is not sleeping but telling you to goe awway!. More exactly, the TCP stack does. Getting a better idea of what's going on on the network layer might be helpful - fire up tcpdump or Wireshark, filter for the database server's address and post the cap to cloudshark.
Sep
3
comment MySQL instance stalling “doing SYNC index”
@MorganTocker I am going to try that in the next maintenance window (a proper DBMS might have allowed for an online change for the transaction log size). The LSN is advancing just ~2.5 M per minute however. If the counter is in bytes (which I suppose it is), the 512 MB of log space available now would take the database through more than 3 hours of operation without a single flush, so I am not all that confident that increasing the log size would make much of a difference.
Sep
2
awarded  Self-Learner