Zbr's days.
←March→
Sun Mon Tue Wed Thu Fri Sat
            1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          
←2008→
Months
Jan Feb Mar
Apr May Jun
Jul Aug Sep
Oct Nov Dec

About :: TODO :: Blog :: RSS :: Old blog :: Projects :: GIT :: Gallery :: Notes

Sun, 30 Mar 2008

Continue DST roadmap.

So, I have to admit that I rethought my opinion about mirroring/redundancy at filesystem layer - it is useful for lots of cases, and modulo bugs in DST mirroring (mostly a leak, which I can not find in my lab, and network/block layer race, which exists in sendfile() for years and just strikes DST a lot, which has a workaround though) I decided to rewrite mirroring algorithm in a way it could be used in other projects.

There is also an idea of how to fix abovementioned network/block layer race in a very non-disturbing manner, which was privately called soft DST barriers. Idea is to replace skb destructor with private one, which will commit that pages are no longer used (for example call bio_endio() or release splice buffer), this callback will be installed only for special sockets, which provide it (like DST, sendfile() or any other ->sendpage() users like samba). Idea was not killed on its roots, which is a good start sign.

/devel/dst :: Link / Comments (8)

Chris wrote at 2008-04-01 17:46:

You know that Symantec has a product feature called DST already, right? Dynamic Storage Tiering? Does a different set of tasks, but it already out there.

Zbr wrote at 2008-04-01 17:55:

Even Linux kernel already has a module called dst - dvb/bt8xx/dst driver, so I'm really late on this name selling auction :)

anonymous wrote at 2008-05-15 08:10:

Are there any updates in the DST roadmap ?

Thanks a lot!

Zbr wrote at 2008-05-15 20:07:

Not much progress so far, sorry. What particular part of the roadmap you are most interested in? :)

anonymous wrote at 2008-05-18 07:45:

Sorry for the delay. I was busy at university and almost spent no time at home so i didnt see this tab in my browser :)

I was intrested in the mirror algorithm mainly. Actually I never tried it, but I am very intrested and hope that after exams I could try it (at work this could be useful, but in the next weeks I am probably going to quit this job and take an other one, so I dont know if I could try this at work with more "real" scenarios)

Zbr wrote at 2008-05-18 12:53:

You can try to use it, it works, syncs and allows node hoplug/hotunplug. I had reports about memory leak likely in resync code, but I can not reproduce it locally, so progress in that area is quite small. Soft DST barriers are not supported either, since my initial proposed network changes were not 100% correct and I did not complete them yet, so you will have to turn scatter/gather capabilities off on your NIC, which does not noticebly degrade perfromance though.

Serge Leschinsky wrote at 2008-06-03 22:40:

Eugene, may I suggest my environment for the memory leak investigation? I'm still very interested in the dst using...

Zbr wrote at 2008-06-04 13:04:

Ok, let's do it new way: next week I will rewrite some bits of DST network core to be esentially like POHMELFS network stack, so it could be done async. Then I will make a new release and we can check if things work smooth.

Please solve this captcha to be allowed to post (need to reload in a minute): 49 * 9

Comments are closed for this story.