Re: [Openqrm-user] nfs server not responding on boot

[Matt R. <ma...@ql...>]

Hey Brandon,

Brandon Allhands wrote:
> Well, you sir, are a genius.

;) thanks for the flowers. Happy when i can help

>
> This worked. 

great to hear !
As i experienced nfs works more reliable over tcp instead of udp anyway.

> What is odd is why does it work for some machines within the same 
> blade center and not others.

yep, that is strange.
Again the question if you are using a single shared fs-image or a bunch 
of "clones"


thanks again + stay tuned,

Matt

>
> I have another full blade center chassis that i am going to be freeing 
> up here in the next few weeks, so I will try again with that chassis 
> and really eliminate if it's hardware or not (Which I am thinking it 
> is now).
>
> The other thing it could be is I do have a cisco load balancer sitting 
> between the QRM nodes and teh nfs servers. I am now thinking it might 
> be messing with udp for some odd reason.
>
> Thanks again for your help!!
>
> Brandon
>
> Matt Rechenburg wrote:
>> Hi again Brandon,
>>
>> ok, another idea. Normally nfs is using udp. Let's try switching it 
>> to tcp.
>> You can do this easily by giving your VE an extra custom, parameter 
>> in the configurtation page.
>> First stop the VE and then put in as extra VE-parameter
>>
>>    NFSMOUNTOPTIONS="tcp"
>>
>> save and then start the VE again.
>> -> this will make the nfs-mount of the rootfs using tcp instead of udp.
>> Maybe it helps to solve the nfs/locking problems.
>>
>> many thanks + all the best,
>>
>> Matt
>>
>>
>> Brandon Allhands wrote:
>>> Yes, there is a gateway, and it is pingable. The NFS server sits on 
>>> the same LAN segment as the blade center (It's big, a /20, but 
>>> that's more for organization than actual number of available ips). I 
>>> have tried swapping all of the services so they run on the same 
>>> subnet, but that didn't do anything.
>>>
>>> Also, nfslock is running, and I have even tried restarting it on the 
>>> server to reset everything in the middle of these timeouts, to no 
>>> avail.
>>>
>>> Brandon
>>>
>>> Matt Rechenburg wrote:
>>>> Hi Brandon,
>>>>
>>>> this situation (as i experienced) can be caused by either not 
>>>> having nfslocking
>>>> enabled+running on the nfs-server or by incorrect dns and/or 
>>>> routing information
>>>> given by the dhcpd-server (options router ...).
>>>> Is the default gateway pingable from the booting system ?
>>>>
>>>> many thanks + have a great day,
>>>>
>>>> Matt
>>>>
>>>>
>>>>
>>>> Brandon Allhands wrote:
>>>>> Hello everyone.
>>>>>
>>>>> Here is what is going on. While booting a QRM node (Which is an 
>>>>> IBM HS20 blade within a chassis), I get the following error after 
>>>>> all of the mounts:
>>>>> nfs: server 192.168.4.11 not responding, still trying.
>>>>>
>>>>> The machine running that IP can ping the machine trying to mount, 
>>>>> and it even shows the authenticated mount requests in the log 
>>>>> right before it supposedly drops.
>>>>>
>>>>> Other machines mount NFS from that IP, even within the same blade 
>>>>> center chassis. This problem is also not limited to 1 machine, 
>>>>> there are several machines in this chassis that are doing this. 
>>>>> All other communication seems to be fine, but without a shell to 
>>>>> get in to when this problem occurs, it severely limits the tools I 
>>>>> can use to test this.
>>>>>
>>>>> Even while this error is occurring, other machines that are 
>>>>> mounted using that NFS server operate fine, with a minimal load on 
>>>>> the NFS server.
>>>>>
>>>>> Has anyone else experienced this, or have any ideas what to try 
>>>>> next? Unfortunately, all of the network connections to the blades 
>>>>> are hardware internal, so I can't stick a transparent bridge with 
>>>>> a sniffer between the blade and the rest of the network (Without 
>>>>> picking up all traffic from the other 13 blades).
>>>>>
>>>>> Thanks,
>>>>>
>>>>> Brandon
>>>>>
>>>>> ------------------------------------------------------------------------- 
>>>>>
>>>>> SF.Net email is sponsored by:
>>>>> Check out the new SourceForge.net Marketplace.
>>>>> It's the best place to buy or sell services
>>>>> for just about anything Open Source.
>>>>> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace 
>>>>>
>>>>> _______________________________________________
>>>>> Openqrm-user mailing list
>>>>> Ope...@li...
>>>>> https://lists.sourceforge.net/lists/listinfo/openqrm-user
>>>>>
>>>>>   
>>>>
>>>>
>>>
>>>
>>
>>
>
>


-- 
www.openQRM.org 
- Keeps your Data-Center Up and Running

Matt's blog - http://mattinaction.blogspot.com/

Please notice my Courses/Workshops for 2008 at the linuxhotel :

openQRM Data-Center Management Plattform
http://www.linuxhotel.de/kurs/openqrm/index.html

Open Source SAN and Cluster-Filesystems
http://www.linuxhotel.de/kurs/san_und_cluster_dateisysteme/index.html