Lessfs-0.7.2 fixes a typo in lib_tc.c that could potentially lead to data corruption.
Please upgrade a.s.a.p. when you use lessfs-0.7.1 with the tokyocabinet data store. Older releases are not effected.
Lessfs-0.7.2 also contains a minor performance enhancement.
Hi Mark
i have made other test with 0.7.1 and 0.7.2 using FileIO and from 0.7beta the performance seems better also in cpu usage, perhaps the 0.7.1 the cpu is little higher but with write performance little much higher compared to 0.7.2.
I dont test using TokioCab. About database tools to check the consistency have you some news?
Many thanks
Dimitri
I am afraid that I haven’t even started on these tools. I have been busy on improving lessfs performance. The next release will support a new hash that will increase performance at least 20% for writes and often much more.
I am still thinking about the file_io backend. It might be necessary to add some information to make integrity checks more easy.
I am using lessfs on ubuntu jaunty 64bit and have noticed a definite slowdown in writing data to lessfs. It is still acceptable for me but I would like to tinker with settings to see if I can improve it. I am basically using the default settings in the provided cfg file and using the example lessfs with big_writes, max_reads, 128k blocks etc. I notice that when I am writing to lessfs that there are noticable pauses when writing large files, these pause almost always occur at increments of 256 MB. I hope to tinker with the settings to correct this. Also, when compiling lessfs, tokyo cabinet and fuse, I did not specify any special settings, just used ./configure, make, make install.
A few questions:
1. If i make changes to settings in the cfg file using the same directories, will that cause data corruption?
2. Can I run mulitple lessfs file systems at once on the same system using different settings in different cfg files.
3. Any reason for the pauses at increments of 256 MB?
I would like to play around with the settings in the config file to find the best for my system but dont want to risk data loss as I have already stuffed many GB of data into lessfs.
Thanks for your hard work in getting this going. This is a great project.
Hi Truss,
The reason why you see delays after writing 256M is that you have probably configured the cache size in lessfs.cfg as 256MB. When lessfs has 256MB in cache it will flush the cache to disk. All other IO will suffer when this happens.
You can of course configure a smaller cache size. The problem will then disappear, however the file systems throughput will go down.
You can change the cache settings safely. You do have to umount/mount lessfs to make the setting active.
You can play with the _bs settings of the databases. To activate these settings you either have to run tc’s optimize or telnet localhost 100 and run defrag.
I would recommend making a backup of the lessfs databases.
Mark.