We are using ZFS on a Sun E450 server (4 x 400 MHz CPU, 1 Gb memory, 18 Gb system disk and 19 x 300 Gb disks running OSOL snv 134) for archive storage where speed is not important. We have 2 RAID-Z1 pools of 8 disks plus one spare disk shared between the two pools and this has apparently worked well since it was set up several months ago.

However, one of our users recently put a 35 Gb tar.gz file on this server and uncompressed it to a 215 Gb tar file. But when he tried to untar it, after about 43 Gb had been extracted we noticed the disk usage reported by df for that ZFS pool wasn't changing much. Using du -sm on the extracted archive directory showed that the size would increase over a period of 30 seconds or so and then suddenly drop back about 50 Mb and start increasing again. In other words it seems to be going into some sort of a loop and all we could do was to kill tar and try again when exactly the same thing happened after 43 Gb had been extracted.

Thinking the tar file could be corrupt, we sucessfully untarred the file on a Linux system (1 Tb disk with plain ext3 filesystem). I suspect my problem may be due to limited memory on this system but are there any other things I should take into consideration? It's not a major problem as the system is intended for storage and users are not supposed to go in and untar huge tarfiles on it as it's not a fast system ;-)


Andy Thomas,
Time Domain Systems

Tel: +44 (0)7866 556626
Fax: +44 (0)20 8372 2582
zfs-discuss mailing list

Reply via email to