181 reputation
7
bio website chrisarmstrong.me
location Seattle, WA
age 27
visits member for 3 years, 7 months
seen Oct 2 at 21:03

Engineering + people + culture. Past lives: engineering leader, ops engineer, product builder. Currently building and evangelizing Deis, an open-source PaaS built on CoreOS and Docker. #hugops


May
7
awarded  Commentator
May
7
comment How to enable the MySQL Slow query log (5.5.15)
@Rick - it's just a boolean on or off. Valid values are 0 or 1.
Sep
26
comment Mitigate HTTP connect floods with HAproxy + Apache
Also, if HAProxy doesn't forward on connections with no requests, I'm afraid that's not what we're dealing with since the app servers have most definitely been getting the requests. So, I guess it makes sense to attempt to mitigate the floods.
Sep
26
comment Mitigate HTTP connect floods with HAproxy + Apache
Hi Willy, thanks so much for the reply! I actually read that article while investigating last night. I set the maxconn directive for each backend server to 40 (It was the original 255 before!). These servers are AWS ec2.large instances, so they have 4 cores. I'm messing around with 1.5dev, but getting it into production will take at least a few days of testing. I saw all of the awesome features it has to protect floods, so I'd like to finish testing ASAP.
Sep
25
revised Mitigate HTTP connect floods with HAproxy + Apache
added 619 characters in body
Sep
25
asked Mitigate HTTP connect floods with HAproxy + Apache
Mar
12
awarded  Scholar
Mar
12
accepted Replacing directories that are subversion-monitored
Mar
12
comment Replacing directories that are subversion-monitored
That sounds like a reasonable enough solution. Thanks!
Mar
12
awarded  Student
Mar
12
asked Replacing directories that are subversion-monitored
Dec
15
comment Few requests/sec, but choke Apache and spike CPU
Thanks for your input, everyone. After looking at the logs once again, it appears that our apache connections per second does indeed spike considerably, to over 100. However, apache requests only spikes to 15-20.
Dec
14
comment Few requests/sec, but choke Apache and spike CPU
I'm thinking rate-limiting at the HAProxy level may not be a bad idea...
Dec
14
comment Few requests/sec, but choke Apache and spike CPU
Thanks for the question. Actually, since we are running a CMS, the invalid requests are getting a 302 redirect to a custom 404 page served by the CMS. Apache is indeed the top CPU user during the spikes.
Dec
14
awarded  Editor
Dec
14
revised Few requests/sec, but choke Apache and spike CPU
added 27 characters in body
Dec
14
asked Few requests/sec, but choke Apache and spike CPU
Sep
21
awarded  Supporter
Sep
9
comment How to enable the MySQL Slow query log (5.5.15)
dev.mysql.com/doc/refman/5.5/en/query-log.html
Sep
9
comment How to enable the MySQL Slow query log (5.5.15)
general_log=1 and general_log_file=/var/log/mysqld.general.log