11,742 reputation
12641
bio website
location Oxford, United Kingdom
age 30
visits member for 3 years, 9 months
seen 15 hours ago

Dec
9
comment Linux kernel bug w/ netns and/or Xen guest - “unregister_netdevice: waiting for lo to become free”
If you down the lo device in the child namespace before tearing it down does that improve things?
Dec
8
comment Getting groups and permission right for apache uploads
If you're making a folder permissions rw, then setfacl -R -m u:ssh2user :rw /home/uploads wont cut it as you need the 'x' perm. Use setfacl -R -m u:ssh2user:rwx /home/uploads instead
Dec
4
comment RPC timeout between 2 linux servers
Provide the result of iptables -vnL
Dec
4
comment Should I sudo as root or use a public key
Nothing stopping you doing both.
Nov
29
awarded  Constituent
Nov
24
comment Server overloaded, acts like out of memory, but thats not true
If the host is still like this now, can you post the results of cat /proc/buddyinfo
Nov
21
comment Server overloaded, acts like out of memory, but thats not true
I think KSM fragmented you. The buddyinfo looks healthy. You want each column to have at least some number but getting lower and lower as you go further to the right. In your case you want to make sure there is at least a few thousand in the first 4 numeric columns.
Nov
21
comment Server overloaded, acts like out of memory, but thats not true
May cause %sys to go up for a few seconds. Expect the host to be unresponsive against a particular CPU for about 20 seconds or so. This wont cause downtime, note the system does this automatically without your consent anyway on certain memory allocations.
Nov
21
answered Server overloaded, acts like out of memory, but thats not true
Nov
21
comment Server overloaded, acts like out of memory, but thats not true
Theres some info missing from the memory dump, can you provide the next 10 lines up from active_anon:79945387 inactive_anon:3660742 isolated_anon:0
Nov
19
answered Evaluating the CPU I/O wait on Linux
Nov
17
awarded  Caucus
Nov
17
comment Keepalived unwanted transition to master
What version of keepalived is it and what distro are you using?
Nov
16
comment What is the size of an IO Operation (IOP) in AWS EBS?
According to this: wiki.archlinux.org/index.php/Advanced_Format "The logical sector size is the sector size used for data transfer", which in your case directly affects your IOP count. Tests I ran on my physical disk with 4k physical / 512 byte logical confirmed the statement. There is no way to alter this parameter as it is advertised by the disk as-is.
Nov
15
awarded  linux
Nov
8
answered Cause of page fragmentation on “large” server with xfs, 20 disks and Ceph
Nov
8
comment Cause of page fragmentation on “large” server with xfs, 20 disks and Ceph
Thats right, but currently theres hardly anything to worry about with regards to fragmentation. I'm writing an 'answer' to try to collate some details together.
Nov
8
comment Cause of page fragmentation on “large” server with xfs, 20 disks and Ceph
Sure, but dropping the caches makes the zone defrag redundant.
Nov
8
comment Cause of page fragmentation on “large” server with xfs, 20 disks and Ceph
/proc/buddyinfo gives you the fragmentation of the page orders. If you manually defragment the zones using echo 1>/proc/sys/vm/compact_memory does that not significantly alter the zone ordering?
Nov
7
comment Cause of page fragmentation on “large” server with xfs, 20 disks and Ceph
Also the output of /proc/meminfo