Starwind 5.6 + Citrix Xen 5.6 FP1 full sync problem

Software-based VM-centric and flash-friendly VM storage + free version

Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)

idigital
Posts: 12
Joined: Sun Feb 27, 2011 5:18 am

Mon Mar 14, 2011 6:57 am

My HA seems to goes out of sync almost every night and when it does, it does a full sync. It is driving me nuts.

What I am seeing in logs:

3/13 23:22:51.135 fe4 HA: CSynchBarrier::LeaveSynchBarrier: WARNING: Barrier block with ThreadId 0xD54 is not found!

Anyone running into this problem?
User avatar
Max (staff)
Staff
Posts: 533
Joined: Tue Apr 20, 2010 9:03 am

Mon Mar 14, 2011 9:38 am

Hello,
This is a known issue, Citrix has confirmed it from their side too.
We're working on a fix right now, will update the thread as soon as I get new information.
Max Kolomyeytsev
StarWind Software
User avatar
Max (staff)
Staff
Posts: 533
Joined: Tue Apr 20, 2010 9:03 am

Mon Mar 14, 2011 4:03 pm

Got the solution!
Here are the working settings:
StarWind Server registry
MaxBurstLength - 0x00040000(262144)
MaxRecvDataSegmentLength 0x00040000(262144)
MaxTransferLength 0x00100000 (1048576)
StarWind Management console:
Max Burst Length 262144
Max Receive Data Segment Length 1048576

XenServer side:
edit iscsid.conf ->
maxburstlength - 65536
maxreceivedatasegment - 131072
Added node.con[0].tcp.window_size = 65536

verify that those are passsed through to etc/iscsi/nodes/iqn/ip/default
Max Kolomyeytsev
StarWind Software
idigital
Posts: 12
Joined: Sun Feb 27, 2011 5:18 am

Mon Mar 14, 2011 8:04 pm

Ah nice,

I already changed the registry which I read about in another post but noticed that the console still had old values. I was about to change the Maxburst from console last night but thought I post this before I mess up my servers :)

I will try this tonight and let you know how it turns out.
idigital
Posts: 12
Joined: Sun Feb 27, 2011 5:18 am

Tue Mar 15, 2011 8:50 am

ok added all the settings and fixed:

node.con[0].tcp.window_size = 65536

to (double n, conn)

node.conn[0].tcp.window_size = 65536

All seems well until 1:09 AM. While it was still doing a sync at 26%, it went out of sync again and now it is at 0% :(
So even the Partner target was not connected to the xenserver's, it went out of sync.

I will wait for the full sync to happen and reconnection to see if this changes. It will take about 24hours thou for a 4TB volume :(

Some observation, for the MaxBurstLength, should this be 80000 instead of 40000 as per post?

http://www.starwindsoftware.com/forums/ ... t2172.html

Tony
Idigital
User avatar
Max (staff)
Staff
Posts: 533
Joined: Tue Apr 20, 2010 9:03 am

Tue Mar 15, 2011 12:46 pm

If this is a test HA volume I'd recommend to remove it and recreate without syncrhonization
As for the values mentioned in the Xen topic before - these block sizes do not work anymore
Max Kolomyeytsev
StarWind Software
idigital
Posts: 12
Joined: Sun Feb 27, 2011 5:18 am

Tue Mar 15, 2011 5:49 pm

Unfortunate I have live data on it.
User avatar
Max (staff)
Staff
Posts: 533
Joined: Tue Apr 20, 2010 9:03 am

Tue Mar 15, 2011 6:10 pm

Well, then the only thing you can do is wait.
I will highly recommend to doublecheck the updates compatibility for future.
Running with Citrix XenServer can be problematic at some points.
Max Kolomyeytsev
StarWind Software
rchisholm
Posts: 63
Joined: Sat Nov 27, 2010 7:38 pm

Tue Mar 15, 2011 7:26 pm

Max (staff) wrote:Well, then the only thing you can do is wait.
I will highly recommend to doublecheck the updates compatibility for future.
Running with Citrix XenServer can be problematic at some points.
Max, when I posted before saying that I was going to be using Citrix Xen Desktop with StarWind and asked if there were any problems with my proposed builds, you didn't say anything about it being problematic. We just brought our new data center up and spent a couple hundred thousand dollars on hardware and software but it isn't going to work correctly when I start the build next week? Or does it mean that I can't upgrade to FP1 yet?
User avatar
Max (staff)
Staff
Posts: 533
Joined: Tue Apr 20, 2010 9:03 am

Tue Mar 15, 2011 7:51 pm

this issue came up suddenly, prior to 5.6 everything was ok.
So I just ask to be careful with XenServer
Max Kolomyeytsev
StarWind Software
rchisholm
Posts: 63
Joined: Sat Nov 27, 2010 7:38 pm

Tue Mar 15, 2011 8:09 pm

Max (staff) wrote:this issue came up suddenly, prior to 5.6 everything was ok.
So I just ask to be careful with XenServer
Didn't 5.6 come out last June? I need some of the features of it. I can do without FP1 though.
User avatar
Max (staff)
Staff
Posts: 533
Joined: Tue Apr 20, 2010 9:03 am

Tue Mar 15, 2011 8:15 pm

StarWind does work with 5.6 fp1 now, we have tested this in our labs, I was talking about doublechecking before updating the environment to avoid the situation of a long sync like in Idigital's case
Max Kolomyeytsev
StarWind Software
rchisholm
Posts: 63
Joined: Sat Nov 27, 2010 7:38 pm

Tue Mar 15, 2011 8:28 pm

Max (staff) wrote:StarWind does work with 5.6 fp1 now, we have tested this in our labs, I was talking about doublechecking before updating the environment to avoid the situation of a long sync like in Idigital's case
That's a load off my mind. This is a new build that will be heavily tested before going into production next month.
User avatar
anton (staff)
Site Admin
Posts: 4021
Joined: Fri Jun 18, 2004 12:03 am
Location: British Virgin Islands
Contact:

Thu Mar 17, 2011 5:28 pm

Deal! Please keep us updated. Thank you!
rchisholm wrote:
Max (staff) wrote:StarWind does work with 5.6 fp1 now, we have tested this in our labs, I was talking about doublechecking before updating the environment to avoid the situation of a long sync like in Idigital's case
That's a load off my mind. This is a new build that will be heavily tested before going into production next month.
Regards,
Anton Kolomyeytsev

Chief Technology Officer & Chief Architect, StarWind Software

Image
idigital
Posts: 12
Joined: Sun Feb 27, 2011 5:18 am

Fri Mar 18, 2011 7:34 am

Just an update, after 2 days operating in full sync mode, the problem has not come back. I think these settings have fixed my problem. If it changes, I'll post an update.

Thanks for all the help!

Tony
Idigital
Locked