Re: [squid-users] Squid Under High Load

From: Chris Robertson <crobertson@dont-contact.us>
Date: Mon, 29 Jan 2007 16:47:00 -0900

Matt wrote:
>> could you convert "large number" and "high load" into decimal numbers?
>
> Learning more as I go here. Installed Cache Manager. See below.
> When it was last started it was completely recompiled and all cache
> files deleted. Am I going to have a problem with file descriptors?

Not so far. If this load is indicative of what your cache is going to
see, you are set. Monitor it during some busy periods just to be sure.

The vmstat output you posted earlier shows a sleeping system. Just
about all of the RAM was "Free" (where Free = cache + buffers + free),
the CPU was always at least 98% idle (with no time spent in wait),
interrupts and context switches looked completely normal.

> Looking at top it appears virtually all the RAM is used up. There is
> 1G of DDR2 of which top says 6M is free and its a single SATA drive.

Linux is like that. Any free RAM is used to cache the file system,
which is why such a low cache_mem setting is recommended in squid.conf.
There is some debate on the best use of RAM (let the OS cache the disk,
or let Squid take it directly), but for your setup (>400 users) the
default is fine.

Just be aware, the rule of thumb of how much RAM squid will use is
available in the FAQ
(http://wiki.squid-cache.org/SquidFaq/SquidMemory#how-much-ram), and is
closely related to how much disk space you give it.

>
> Thanks.
>
> Matt
>
>
> --------------------------------------------------------------------------------
>
>
> Squid Object Cache: Version 2.6.STABLE9
> Start Time: Sat, 27 Jan 2007 03:11:47 GMT
> Current Time: Sat, 27 Jan 2007 16:50:51 GMT
>
> Connection information for squid:
> Number of clients accessing cache: 385
> Number of HTTP requests received: 567457
> Number of ICP messages received: 0
> Number of ICP messages sent: 0
> Number of queued ICP replies: 0
> Request failure ratio: 0.00
> Average HTTP requests per minute since start: 692.8
> Average ICP messages per minute since start: 0.0
> Select loop called: 22474678 times, 2.187 ms avg
> Cache information for squid:
> Request Hit Ratios: 5min: 29.7%, 60min: 26.5%
> Byte Hit Ratios: 5min: 10.2%, 60min: 8.0%
> Request Memory Hit Ratios: 5min: 6.3%, 60min: 6.2%
> Request Disk Hit Ratios: 5min: 39.8%, 60min: 37.9%
> Storage Swap size: 2012780 KB
> Storage Mem size: 8172 KB
> Mean Object Size: 16.02 KB
> Requests given to unlinkd: 0
> Median Service Times (seconds) 5 min 60 min:
> HTTP Requests (All): 0.10281 0.10857
> Cache Misses: 0.13498 0.14252
> Cache Hits: 0.00463 0.00463
> Near Hits: 0.17711 0.12106
> Not-Modified Replies: 0.00179 0.00286
> DNS Lookups: 0.07284 0.07284
> ICP Queries: 0.00000 0.00000
> Resource usage for squid:
> UP Time: 49143.658 seconds
> CPU Time: 285.434 seconds
> CPU Usage: 0.58%
> CPU Usage, 5 minute avg: 1.48%
> CPU Usage, 60 minute avg: 1.23%
> Process Data Segment Size via sbrk(): 46168 KB
> Maximum Resident Size: 0 KB
> Page faults with physical i/o: 0
> Memory usage for squid via mallinfo():
> Total space in arena: 46168 KB
> Ordinary blocks: 28736 KB 188 blks
> Small blocks: 0 KB 0 blks
> Holding blocks: 1260 KB 2 blks
> Free Small blocks: 0 KB
> Free Ordinary blocks: 17431 KB
> Total in use: 29996 KB 63%
> Total free: 17431 KB 37%
> Total size: 47428 KB
> Memory accounted for:
> Total accounted: 23071 KB
> memPoolAlloc calls: 73981527
> memPoolFree calls: 73554543
> File descriptor usage for squid:
> Maximum number of file descriptors: 1024
> Largest file desc currently in use: 798
> Number of file desc currently in use: 597
> Files queued for open: 0
> Available number of file descriptors: 427
> Reserved number of file descriptors: 100
> Store Disk files open: 2
> IO loop method: epoll
> Internal Data Structures:
> 125764 StoreEntries
> 1812 StoreEntries with MemObjects
> 1791 Hot Object Cache Items
> 125660 on-disk objects

Chris
Received on Mon Jan 29 2007 - 18:47:11 MST

This archive was generated by hypermail pre-2.1.9 : Thu Feb 01 2007 - 12:00:01 MST