3,101 reputation
218
bio website bo.lt
location SF, CA
age 34
visits member for 2 years, 11 months
seen Jul 20 at 16:54

I love uptime, beer and motorcycles.


Oct
11
comment How to change back /etc/sudoers file right to 0440?
Just for future reference you should use visudo to edit that file: serverfault.com/questions/26303/…
Oct
11
comment LINUX Load average
Can you update your question with what diagnostics you've already done? ps output? iostat output? actual top output?
Oct
11
comment Fixing window scaling problems with Linux TCP
Are you seeing the window shrink to zero during transfers? Are you actually seeing packet delays while one side is waiting for window availability? It'd be helpful if you could post the pcaps (with timestamps) that you think confirm this so they can be analyzed by the community.
Oct
11
comment Postfix development server - intercept all outgoing mail
A lot of these are great ideas, but they all seem to miss the one dev machine that wasn't configured properly and just started sending mail. It is usually a good idea to take care of this by either firewalling or transparently proxying all outbound port 25 traffic on the developer networks.
Oct
10
comment Request from 10.x.x.x a few seconds after original request
@Mischa your server. 10/8 space is not routable over the internet (except through a VPN).
Oct
10
comment Would OOM_Killer succeed if SIGKILL didn't?
SIGKILL didn't fail, the process went to exit and was stuck trying to retrieve all its pages from swap. You could verify this by checking the disk busy and watching swap usage page activity while it was in D. Also by seeing what its state is in top or ps axwwule.
Oct
10
comment Determine if TCP socket is in kernel-space
Can you clarify what you are trying to do? Or what the motivation is here to hopefully clarify what the question is? What information are you trying to gather? What do you mean by 'TCP socket is in kernel-space'?
Oct
10
comment EC2 disk goes to 100% util, but fine after reboot
Can you update with iops r/w counts then? iostat -x 1 output for several seconds perhaps?
Oct
10
comment Apache with prefork model using 1000 MB memory per process
Yes, mod_php and not using fastcgi: fastcgi.com/drupal/node/5?q=node/10
Oct
9
comment EC2 disk goes to 100% util, but fine after reboot
100% disk usage as a statistic itself is almost completely meaningless. Are you tracking number of IOPS (reads and writes), the average request size and svctime for those operations? The next step for solving this is to determine if your application is causing a significant increase in # or size of operations (your issue), or if the operations are starting to take longer (amazon issue) when you see the problem.
Oct
9
answered Apache with prefork model using 1000 MB memory per process
Oct
8
comment What is the benefit to having a network's HTTP traffic sent out on a different IP than HTTPS?
This is a guess at how they are doing it, the question was what are the possible benefits.
Oct
8
comment What is the benefit to having a network's HTTP traffic sent out on a different IP than HTTPS?
This is a guess at how they are doing it, the question was what are the possible benefits.
Oct
8
answered What is the benefit to having a network's HTTP traffic sent out on a different IP than HTTPS?
Oct
8
comment Traffic shaping for gateway to stop leechers
It'll be pretty difficult and unmaintainable to track youtube by IP. You should really be using delay_parameters with squid and URL based classification for that. Youtube has two ASNs (not including google inc's many ASNs) and from just one of them they advertise more than 20 blocks: bgpctl show ip bgp as 36040 | awk '{ print $2 }' | sort | uniq | wc -l 28
Oct
8
comment How are Java Libraries on Debian supposed to work?
We also don't use the system manager for our JVMs and the related libraries/tools. The other thing I ran into, which it looks like you have too is that the package maintainer has not fully tested features and has improperly set the class path, while it isn't hard to change it almost always meant having a custom catalina.sh or activemq-start.sh that meant we were already deviating from the systems packages.
Oct
8
comment providing logs to the users
It doesn't really support distributed backends so its fairly limited, but you can do hundreds of millions of events in a single system pretty easily.
Oct
6
answered How passively monitor for tcp packet loss? (Linux)
Oct
6
comment Easy way to “edit” the traffic coming from a tcp host (linux)
Are these HTTP requests?
Oct
5
comment Initializing squid stops
You should probably uncomment the cache_dir line.