Nyamul Hassan wrote:
Hi,
I would build with the following in mind:
1. Better to have a separate disk for the cache stores.
2. Have a COSS store for objects less than 256k. And let AUFS handle
larger objects.
3. Don't have more than 75% of your disk allocated.
4. Only one AUFS store per disk. But you can have more than one COSS.
Use AUFS on Linux and DISKD on BSD.
5. RAID is highly discouraged.
RAID is not highly discouraged. All depends on functional and technical
requirements. For example, RAID is a highly encouraged if the Squid server
is considered a critical business system. See
http://wiki.squid-cache.org/SquidFaq/RAID.
6. Regex is best avoided
If you want to use regex on URLs to filter access, consider a URL filter.
7. use noatime and nodiratime mount options
8 Specific for Linux:
Add to /etc/sysctl.conf the following lines:
# swappiness can have a value of between 0 and 100
# swappiness=0 tells the kernel to avoid swapping processes out of physical
# memory for as long as possible
# swappiness=100 tells the kernel to aggressively swap processes out of
# physical memory and move them to swap cache
# default: 60
# For an application server we use a less aggressive setting of 20.
vm.swappiness=20
# VFS cache pressure:
# default: 100
# With values lower than 100, the data cache is reduced more and the
# inode cache preserved
vm.vfs_cache_pressure=50
# Overcommit or not memory allocations:
# 2 Don't overcommit. The total address space commit
# the system is not permitted to exceed swap + a
# percentage (default is 50) of physical RAM.
# on the percentage you use, in most situations
# means a process will not be killed while accessing
# but will receive errors on memory allocation as
# appropriate.
# Also ensure that the size of swap is at least 50%
# of physical RAM with a minimum of 2 GB.
vm.overcommit_memory=2
Run the following command to make the new configuration active:
# sysctl -p /etc/sysctl.conf
9. adding disk was already mentioned but may not be omitted in a list.
Marcus
Regards,
HASSAN
On 2010-08-07, Mike Diggins <mike.diggins@xxxxxxxxxxx> wrote:
I'm setting up a RHEL 5.3 64-bit Squid installation using
squid-2.7.STABLE9. It's not a high volume server, and there will be two,
but I wondered if there was anything special I should do with the cache
disk, which is an EXT3 file system. I do mount it with the noatime option.
I'm using AUFS with this cache_dir configuration:
cache_dir aufs /usr/local/squid/var/cache 20000 16 256
My file system is mounted:
LABEL=/cache /usr/local/squid/var/cache ext3
rw,noatime 1 2
I'm not sure the model of the disk, but it's:
73GB 15K RPM SA SCSI 6Gbps 2.5in (Dell system)
Any suggestions (besides adding more disk!).
-Mike
