ZFS_DEBUG_MODIFY and ECC

[dboy]

This is almost a double post, but the earlier discussion is hidden in a thread about user experiences so I think it warrants a proper discussion here.
I have been reading up on ZFS and non-ECC vs ECC in relation to a disk upgrade after a near crash, and acquisition of a second hand old professional rack server.

My conclusion is that non-ECC probably works well, but ECC is better...
I read this recent thread,
viewtopic.php?f=4&t=6244#p68136
and read in a link that one of the ZFS creators, Matthew Ahrens, recommends to turn on the flag
ZFS_DEBUG_MODIFY
in this link
http://blog.codinghorror.com/to-ecc-or-not-to-ecc/

This flag, according to Ahrens, reduce the risk of memory error by checksumming data in memory before commiting a write, if I understand it correctly.

Questions/Suggestions
1.
My question for those who understand the fineprint details in ZFS, is this useful if you have ECC also?
Does it increase safety or maybe even reduce it?
I think ECC commonly correct single bit errors but not double or multiple. Would the flag help with that?

2.
Since Ahrens recommend it, and most NAS4FREE users probably don't have ECC, why not have it turned on as default?

Just to cover why: (Allthough the discussion about ZFS is interesting anyway)
I understand that some nas4free users just keep a copy of media on their servers, but I also keep a large volume of private data, films, photos. I do keep backups, but if there are memory induced errors, they are very likely to propagate through to backups. It is near impossible to check all films and photos to see which copy is unaffected by bit errors that make parts of movies or photos unviewable.

[raulfg3]

This flag, according to Ahrens, reduce the risk of memory error by checksumming data in memory before commiting a write, if I understand it correctly

yes, but is only usefull if you use non-ECC RAM, if you use ECC RAM is not recomended.

Since Ahrens recommend it, and most NAS4FREE users probably don't have ECC, why not have it turned on as default?

Because is a penalty on system resources and your NAS can be slow, so is good to know, but you need to decide (safety Vs Speed).

PD: I suggesto to do some test in your NAS and decide if you have non-ECC RAN if is good for you.


In my home NAS I have non_ECC RAM and i live with it because I only store films and pictures and a small bit rot is not noticeable, but of course is not the same for a BIG DATA CLOUD or service provider. <- This scenario allways use ECC RAM

[dboy]

Hi Raul,
Thanks, I just got a second hand old Supermicro server with ECC, and plan to get 16GB. will equip it with 4*5TB in a RaidZ1. That will be ok for many years for me I hope.

This means I can forget about the debug_modify flag.

However, it is still interesting.
Have you tried setting it?
How bad is the performance penalty?

I know your reasoning from earlier posts, and I see your view. I still prefer to keep things as safe as possible.

[raulfg3]

ow bad is the performance penalty?

in my mobo, about 25% slow (75MBps when I reached 115MBps)