#9460: Many Maxima-related doctest failures on sage.math
-----------------------+----------------------------------------------------
Reporter: mpatel | Owner: mvngu
Type: defect | Status: new
Priority: major | Milestone: sage-4.5.3
Component: doctest | Keywords:
Author: | Upstream: N/A
Reviewer: | Merged:
Work_issues: |
-----------------------+----------------------------------------------------
Comment(by drkirkby):
Replying to [comment:43 mpatel]:
> Replying to [comment:42 drkirkby]:
> > Does this still need to be open?
>
> I still get the same types of failures with a 4.5.alpha4 I compiled
today with the [comment:2 go script above] on sage.math. Is the ZIL is
still disabled on the Sage cluster? I don't know if the plan is to enable
it permanently, but it might help to do it temporarily, if it's practical,
and revisit this ticket and #9501.
Here's the situation.
[http://en.wikipedia.org/wiki/ZFS ZFS] is the file system used on the main
server disk.math. The ZFS Intent Log (ZIL) was disabled by William long
ago (> 1 year). It speeds up NFS writes considerably, but it risks data
corruption on the NFS clients (sage.math, t2.math, boxen.math etc). IMHO,
this is a very bad idea.
William has three choices
* Leave the ZIL disabled and risk data corruption.
* Re-enable the ZIL, get valid data, but at a cost of a dramatic slow
down in NFS speed.
* Buy a fast solid state disk. Then configure the storage pool so the ZIL
is written to the fast solid state disk. The disk does not need to be
large (even 100 MB would be sufficient), but it needs to be a good quality
enterprise grade disks. Logging to a USB memory stick would not be a good
idea.
I've made William aware of this long ago. What he does is up to him. As
far as I'm aware, the ZIL is disabled. Therefore, I would not trust any
file in the home directories at all. I would only trust the disks locally
mounted on the machines. If the problem goes away when things are built in
Looking on sage.math, I see {{{/mnt/usb1/scratch}}} is locally mounted, so
that should not suffer the problems the NFS mounted directories have. (I'm
a bit suspicious that {{{/mnt/usb1 }}} might actually be a USB mounted
hard drive, which undoubtedly uses a consumer grade disk. The disks on a
server like sage.math should not be on USB connectors, which is what that
device name implies to me.)
{{{/scratch}}} on 't2.math' is a high quality local disk, so I trust that
as much as you can trust any single hard drive. It is not backed up and
its not mirrored.
> According to [comment:ticket:8731:40 this comment] at #8731, there's now
a newer upstream release of Maxima. I don't know if it will help here.
I've no idea.
We did try updating both ECL and Maxima recently, and it all went pear
shaped. I don't think that has been resolved. I've rather lost track of
what happened over that.
Dave
--
Ticket URL: <http://trac.sagemath.org/sage_trac/ticket/9460#comment:44>
Sage <http://www.sagemath.org>
Sage: Creating a Viable Open Source Alternative to Magma, Maple, Mathematica,
and MATLAB
--
You received this message because you are subscribed to the Google Groups
"sage-trac" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to
[email protected].
For more options, visit this group at
http://groups.google.com/group/sage-trac?hl=en.
