Reputation
16,388
Top tag
Next privilege 20,000 Rep.
Access 'trusted user' tools
Badges
1 14 30
Impact
~818k people reached

Jun
11
comment How can I achieve ACID on multiple physical front-end servers
@BaronSchwartz: although contract based replication is a lot more robust than async replication, I am not aware of it offering any guarantees that a write completes, only that the request has been accepted and will be run by the target node. Hence far from ACID but enough to meet the requirements of most applications. But happy to be proved wrong if you have any hrefs
May
29
awarded  Notable Question
May
12
awarded  Nice Answer
May
1
comment 99% disk IO spikes with Percona
4 errors on disk 2 - but such a low number is nothing to worry about - certainly nothing to do with the problem described in the question.
May
1
comment 99% disk IO spikes with Percona
You should be checking pretty much every option - lookahead, 32 bit IO mode, DMA mode, multiple IO count, accoustic mgmt mode. Also check for errors with smartctl. But I'd start with checking the disk controller.
Apr
30
comment 99% disk IO spikes with Percona
If its handling exactly the same workload as other servers which are not exhibiting the problem then this rather points to a problem at the hardware level. Is there a caching disk controller? With a flat battery? Is hdparm reporting the expecting config (or paste it here)?
Apr
10
answered nginx + PHP-FPM disable PHP functions only for certain virtual hosts
Apr
7
comment Current setup maxing CPU. Need help designing AWS architecture and MySQL Clustering via AWS RDS instances
There can be big performance issues with Master/slave. Multi-master replication is nothing like web balancing (the whole point is that the DB updates are sent to every node). You will run into extreme pain if you try to run more than 2 masters in an asynchronous cluster. Using memcached to cache results has limited benefit over using the SQL cahce in MySQL - and you will have worse concurrency issues. This is not a good answer.
Apr
1
answered Do I have to load Zend Apache/PHP extensions if I'm not using Zend Framework?
Mar
23
answered Blocking PHP mailer spam using iptables
Mar
22
answered How do I set up a geographic mirror site for LAMP?
Mar
20
answered How to replace a hardware modem by virtual modem
Mar
19
answered 1 IP to multiple different physical web server / web server machines
Mar
19
comment 1 IP to multiple different physical web server / web server machines
I don't believe this question belongs elsewhere and its disappointing to see so many down votes without any comments
Mar
17
comment Should I 'run in' one disk of a new RAID 1 pair to decrease the chance of a similar failure time?
Agree for conventional HDS, but for ssds its a very different story. Thought this was worth noting now before 4tb ssds become cheap and available and readers don't realise we're talking about spinning rust here, but maybe by then they'll handle more writes.
Mar
17
accepted Forward proxy for webservices with SSL client certs
Mar
17
answered Forward proxy for webservices with SSL client certs
Mar
16
comment Preferred method for allowing unprivileged UNIX/Linux users to view syslog information
@0xC0000022L: because its easy to prove that Unix permissions are correct. It's really hard to audit ACLs, and I've seen many ACL configurations where (a deep, manual examination has revealled) there are gaping security holes. And coincidentlally there are also performance issues on some filesystems and issues with backup software support.
Mar
15
comment Forward proxy for webservices with SSL client certs
How does this tell Squid how to select a client certificate specific to the target webservice?
Mar
8
comment email sending speed - how to improve
No, that is not what it means at all. It means that there was at least one io request outstanding for a while when the scheduler had no tasks which were runnable. Nothing is "stalled" based on the metrics provided.