Lustre discuss - Jul 2007 - State information

Hi,

I am thinking of writing a small script that would poll the various  
OSS servers for their states. I was thinking of starting at /proc/sys/ 
lnet/peers to find out who is connected and their states. When i  
printed out the content of "/proc/sys/lnet/peers" on the MGS node  
this is what i see

[root@mgs-1 ~]# cat /proc/sys/lnet/peers
nid                      refs state   max   rtr   min    tx   min queue
0@lo                        1    up     0     0     0     0     0 0
192.168.1.6@tcp             1  down     8     8     8     8     7  
0		// This is OSS-1		
192.168.1.8@tcp             1  down     8     8     8     8     7  
0		// This is OSS-2
192.168.1.10@tcp            1    up     8     8     8     8     7  
0		// This is MDS-1
192.168.1.12@tcp            1    up     8     8     8     8     7  
0		// This is Client-1

The question i have is that the OSS nodes are running and are up but  
on MGS it shows them as down, is there a reason why the state for the  
OSS nodes is down?

Is there some other place on MGS node from where i can get a list of  
nodes in a Lustre Cluster? Since that''s what MGS saves on a MGT is  
there a way to extract that information from the MGS?

Thanks

Anand

There''s a list of the servers participating in each filesystem here,
but
it doesn''t give you current status.
cfs21:/tmp# cat /proc/fs/lustre/mgs/MGS/live/*
fsname: lustre
flags: 0x0     gen: 26
lustre-MDT0000
lustre-OST0000
lustre-OST0001

You can get the MDT''s concept of which OST''s are active here:
cfs21:/tmp# cat /proc/fs/lustre/lov/lustre-mdtlov/target_obd
0: lustre-OST0000_UUID ACTIVE
1: lustre-OST0001_UUID ACTIVE

But there isn''t currently a nice centralized location with the status
of
all the nodes in the cluster.

Anand Bisen wrote:
>
> Hi,
>
> I am thinking of writing a small script that would poll the various 
> OSS servers for their states. I was thinking of starting at 
> /proc/sys/lnet/peers to find out who is connected and their states. 
> When i printed out the content of "/proc/sys/lnet/peers" on the
MGS
> node this is what i see
>
> [root@mgs-1 ~]# cat /proc/sys/lnet/peers
> nid                      refs state   max   rtr   min    tx   min queue
> 0@lo                        1    up     0     0     0     0     0 0
> 192.168.1.6@tcp             1  down     8     8     8     8     7 
> 0        // This is OSS-1       
> 192.168.1.8@tcp             1  down     8     8     8     8     7 
> 0        // This is OSS-2
> 192.168.1.10@tcp            1    up     8     8     8     8     7 
> 0        // This is MDS-1
> 192.168.1.12@tcp            1    up     8     8     8     8     7 
> 0        // This is Client-1
>
> The question i have is that the OSS nodes are running and are up but 
> on MGS it shows them as down, is there a reason why the state for the 
> OSS nodes is down?
>
> Is there some other place on MGS node from where i can get a list of 
> nodes in a Lustre Cluster? Since that''s what MGS saves on a MGT is
> there a way to extract that information from the MGS?
>
> Thanks
>
> Anand
>
>
>
> _______________________________________________
> Lustre-discuss mailing list
> Lustre-discuss@clusterfs.com
> https://mail.clusterfs.com/mailman/listinfo/lustre-discuss