Hello all, I recently setup Squid (2.5.STABLE6) on a dual-cpu Xeon system with 2GB of RAM and 4GB swap. The system was having stability issues before, which I finally attributed to the hardware, namely the disks or the disk controller. The drives showed bad sectors on some badblock tests, though not every time.
For this new installation, I got rid of all the old drives and added two brand new 72GB drives (Seagate, UW SCSI). I created two cache partitions of 48GB each (with identical cylinder values) and for each drive, set 'cache_dir' to 39000 and filesystem to aufs. Once the cache filled up, I was expecting to find out if the same hardware issues appeared, but it seems to have stopped growing at about 70200MB (~68GB). We get an average number of 55 requests per second and these go upto about 120 req/s in peak times so low traffic is not the issue. Is there anything I'm missing that's causing the cache size to stay the same? cache.log doesn't show anything unusual. Nor does the system log. Performance is pretty good too. I'm just worried that the cache might just start to grow when we least expect it and cause the same problems. I'm guessing that Squid is reserving the extra space (~8GB) for internal use, but would like some informative feedback. Thank you in advance. -- A. Sajjad Zaidi http://www.sajjadzaidi.com/ GnuPG Key ID: 0xD7AD0E13 "What's blue and square? An orange in disguise..."
