Topic: Problems when testing 8b6 on new hardware

[lloydh]

I have new server hardware I have been testing, it is not from the certified list because it is a home server I cannot afford commercial server hardware, the selection of the hardware was the subject of a post I made recently in the SME 7 forum.     The hardware is :-

ASUS TA-9A1 VENTO Midi Tower Front 4xUSB 1x FireWire and Audio 450watt PSU
Gigabyte GA-880GM-USB3 AMD SB710 4xDDR3 VGA GLAN 6xSATA2 RAID USB3.0 MicroATX
AMD Phenom II 550 DualCore 3.1GHz 7MB CACHE 80W
Kingston 4GB(2 X 2GB) DDR3-1333MHZ KVR1333D3N9K2/4G
2 x Western Digital 500 gig 7200rpm SATAII DS GREEN POWER 32meg Cache
LG 22x DVD+/-R Burner SATA Model: GH22NS50 DUAL LAYER

This hardware is to replace my existing hardware running SME 7.5.1, I reailsed I took a risk getting the latest technology which could have problems with 7.5.1, the MB has the latest BIOS.      My initial testing was on 7.5.1, I did have problems, I have found workarounds for these problems but the workarounds seem to impact on performance but I can live with that until 8.0 is released.       I am now testing SME 8.0beta6 and I have some questions and issues, it is configured as server only for testing and has been restorred from the backups from my 7.5.1 server and gateway server, I obviously changed the IP and name after the restore as both servers are on the same network.

Question 1.    My old server does a backup to a USB disk using the DAR backups built into Server Manager, the server is slow and a full backup takes longer than 24 hours so I time out the backups at 23 hours and inc1 picks up the rest of the full backup the next day.       My question is when I do a full restore, if I pick the backup set and then say select inc3, am I right in assuming this will restore inc3 + inc2 + inc1 + full which is what appears to happen.

Question 2.    The handling of mountable drives in 8 is different to what happens in 7, what I mean is in 7 my fstab has mount points for the CD-ROM, floppy and USB disk which are added during boot, in 8 this does not happen and I had to add the entry and cerate the directory in /media so I could use my USB disk.      Is this the way it will be in 8, if so, will there be documentation to say what to do.      In the header of fstab it says to look at http://www.e-smith.org/custom/ which is now a dead link or look at the template fragments in /etc/e-smith/templates/etc/fstab/ but I don't know what I am supposed to do with these.    I know there are USB issues with 8 at the moment, is this related.

Issue 1.    I have the data from my 7.5.1 server on the 8 server and I run the same USB backups, a full weekly backup followed by incrementals with 3 sets.      Things work fine then I get this email randomly.

==================================
DAILY BACKUP TO WORKSTATION REPORT
==================================
Backup of wombat started at Thu Oct 21 11:25:02 2010
Backup of mysql databases has been done
Mounting backup shared directory localhost/media/wd001
*** No backup allowed or error during backup ***
Error while mounting  /media/wd001 :
mount: can't find /media/wd001 in /etc/fstab or /etc/mtab

The reason for the error is that it is looking for /media/wd001 but it should be looking for /media/WD001 which it has foung during the previous backups, I also noticed that there is now a /media/wd001 directory.        If I go into Server Manager > Backup or restore > Configure workstation backup it also shows /media/wd001.      If I now go Next the Backup share in the drop down window is still showing as WD001, if I go Update without changing anything and then look in Configure workstation backup again it is now /media/WD001 and the backups work again for a few more days before it happens again.

Issue 2.    During boot it displays "TSC appears to be running slowly. Marking it as unstable" and pauses for a couple of seconds.     Using Google I find this is related to the clock but I don't have any problems with the clock or anything else and everything seems to be stable, is this something to worry about.      I notice that in the logs before I get the previous message there is "Time: hpet clocksource has been installed.".

I am relative new to these forums even though I have been using SME Server for years, what I am looking for is guidance.      If you think I need to raise any bugs please tell me and I will do that.

Thanks, Lloyd.

[janet]

lloydh

A1) Yes

[janet]

lloydh

Issue 1......The reason for the error is that it is looking for /media/wd001 but it should be looking for /media/WD001 which it has found during the previous backups...

I suggest you lodge a bug report about this. It does relate to beta release software, so there may be an issue there.

[janet]

lloydh

Issue 2.    During boot it displays "TSC appears to be running slowly. Marking it as unstable" and pauses for a couple of seconds.

   

Looking at the google search results and guessing,
possibly this http://isp-control.net/forum/thread-2075-post-19476.html#pid19476
or
http://forums.fedoraforum.org/showthread.php?t=154867
or this (which suggests the issue is kernel version related)
http://forum.proxmox.com/threads/3238-TSC-appears-to-be-running-slowly.

Seems to be hardware & kernel related, so maybe search more at Gigabyte site for switches and/or fixes.

[janet]

lloydh

A2) Refer to
http://bugs.contribs.org/show_bug.cgi?id=6271
which refers to
http://wiki.contribs.org/USBDisks#USB_on_SME_8

If you are going to use and play with SME8beta releases, then you need to follow the development process and bug fixes and read the SME8 wiki article for latest news and changes etc. After all it is still being developed so changes will & do occur.

[lloydh]

Mary,

Thanks for your effort, I have some personal issues that prevent me from spending a lot of time on this at the moment so there may be delays in getting things done but I do want to see this through.

You have given me a lot to read and I will also create a bug for issue 1.       I haven't found much on the Gigabyte site but the following statement on their site doesn't give much encouragement "Due to different Linux support condition provided by chipset vendors, please download Linux driver from chipset vendors' website or 3rd party website".

As for issue 2, I forgot to mention at one stage I had loaded Centos 5.5 when trying to resolve another issue and I had kept the Centos log and there are no TSC errors in the log.       Also I have obviously kept 8.0b6 up to date with the latest patches.

[janet]

lloydh

Gigabyte GA-880GM-USB3 AMD SB710
I am now testing SME 8.0beta6

Motherboard compatibility with Linux should always be checked before purchase.
Your supplier is usually a good person to ask.
That board appears more suited to Windows 7/Vista/ XP

Having said that, if Centos 5.5 ran OK, then SME8 should run OK, as SME Server 8.0beta6 is based on CentOS 5.5. You may want/need to disable some aspects of the motherboard functionality in the BIOS settings in order to improve compatibility.

Later releases of SME8.x may have a newer kernel & drivers and therefore support your m/b features more fully.

[lloydh]

Motherboard compatibility with Linux should always be checked before purchase.
Your supplier is usually a good person to ask.
That board appears more suited to Windows 7/Vista/ XP

Having said that, if Centos 5.5 ran OK, then SME8 should run OK, as SME Server 8.0beta6 is based on CentOS 5.5. You may want/need to disable some aspects of the motherboard functionality in the BIOS settings in order to improve compatibility.

Later releases of SME8.x may have a newer kernel & drivers and therefore support your m/b features more fully.

Mary, I have used Asus motherboards in the past but changed to Gigabyte after advice from more than one source that stated Gigabyte boards are more linux friendly than Asus motherboards.     In the BIOS I have disabled all hardware I don't need such as Firewire, USB3, Audio, Serial, Floppy etc.

The reason behind trying Centos was if the problem existed there I would have raised a bug with Centos, again the TSC problem doesn't seem to have any impact on the server performance which is good and the clock is stable.        Re your links about TSC, yes, I did a lot of searches in Google and came up with similar information and I have tried many "clock" options as kernel parameters with no luck, I was using http://www.mjmwired.net/kernel/Documentation/kernel-parameters.txt as a reference.

Some extra information relating to Issue 2, when I got the motherboard it had F6 BIOS and I immediately upgraded to the latest which was F7 and then installed 7.5.1.     Everything worked fine except the clock was running at double speed, after many Google searches I found if I booted in UP mode with NOAPIC as a kernel parameter the clock was stable but the performance wasn't good but I can live with that until the final release of 8.0.       I then tried 8.0 beta 6 and as well as the TSC error I got the following.

Oct 11 16:12:30 wombat kernel: APIC error on CPU1: 40(40)
Oct 11 16:13:17 wombat kernel: APIC error on CPU1: 40(40)
Oct 11 16:14:32 wombat kernel: APIC error on CPU1: 40(40)
Oct 11 16:16:05 wombat last message repeated 5 times

Again I couldn't resolve this with Google searches and kernel parameters but in late October I noticed F8 BIOS had been released for the motherboard and I installed that and noticed immediately the APIC CPU errors were gone.       Having thought about that for a while I decided to raise a fault ticket with Gigabyte regarding the TSC problem and I will wait and see what happens with that.

Regarding the Issue 1 backup problems, I did some more testing with that and eventually found it always happened after a software update and I can now simulate it by just performing a signal-event post-upgrade and no reboot and I have opened bug 6387 to report it.

Lloyd.

[janet]

lloydh

Regarding the Issue 1 backup problems..... I can now simulate it by just performing a signal-event post-upgrade and no reboot and I have opened bug 6387 to report it.

Are you saying in that bug report, that in the SME Server Manager Backup or Restore panel, somehow (without direct user intervention) the drive location changes from /dev/WD001 to /dev/wd001 when a post-upgrade event is done ?

[lloydh]

Are you saying in that bug report, that in the SME Server Manager Backup or Restore panel, somehow (without direct user intervention) the drive location changes from /dev/WD001 to /dev/wd001 when a post-upgrade event is done ?

Mary, no I am saying it is changing /media/WD001 to /media/wd001 in the Backup and restore panel.       To me this must be something in some configuration file and it has actually created a /media/wd001 directory as well, the creation time is the backup start time.      If I check the device label using HAL before and after the post-upgrade it is always WD001, the device is not being changed, just the path to the mount point in the backups.

[janet]

lloydh

Oops sorry, I meant to say changing the setting of /media/WD001 to /media/wd001 in the Backup and restore panel.       

Would/could this be a more simple issue of case, the upper case is not being retained by SME which is somehow changing it to lower case.
It would be interesting to see if you can change the label on the USB drive to lower case ie wd001, reconfigure the backup panel, and then see if all goes well through post-upgrade changes etc.

The ftsab entry seems correct, it is a managed entry so the system is identifying the drive correctly as /media/WD001, it's just that it is changing case in the backup panel entry.

Check the db entries before and after the post-upgrade command is issued.
config show backupwk
You should also be able to scour the messages log file around the time of doing the change, to see where the entry is made.

Actually all this stuff should go in the bug report.

[CharlieBrady]

If you think I need to raise any bugs please tell me and I will do that.

Anything that doesn't work perfectly should be reported as a bug.

[chris burnat]

A bug has ben lodged  in Bugzilla and this issue has been resolved.  Thanks for reporting lloydh.
http://bugs.contribs.org/show_bug.cgi?id=6387

[CharlieBrady]

A bug has ben lodged  in Bugzilla and this issue has been resolved.  Thanks for reporting lloydh.
http://bugs.contribs.org/show_bug.cgi?id=6387

lloydh appears to be reporting multiple problems in this thread. I think that only one has been reported and then fixed in bugzilla.

[lloydh]

Anything that doesn't work perfectly should be reported as a bug.

Point taken Charlie, I am new here and thought the forums would be used to determine if there was a bug then I open a bug report and I apologise for putting multiple issues in the same post.

Question 1 has been answered.

Question 2 has been answered.

Issue 1 has been resolved through the bug tracker.

Issue 2 is still there but doesn't appear to be causing any problems, what I will do is install Centos 5.5 again and carefully examine the logs, if the TSC problem is not there in Centos I will raise a new bug.

[mmaas70]

Just wanted to say that although reporting bug in the tracker is useful, this thread helped me identifying the same upper/lower-case usb problem. Glad that's solved now, and thanks to the problemsolvers of this great, great distro...

[lloydh]

Issue 2 has been raised as Bug 6466 asking if this is an error or just a message that can be ignored.