September 30, 2012
I have 11.2 Oracle cluster installed on two HPUX 11.31 ITANIUM nodes. After some years of work one node failed to mount NFS NettAPP storage share, second node have no issues. On first node mount hangs.
inetapp:/ora_voting_disk/hpuxia /orarac nfs rw,bg,vers=3,proto=tcp,noac,forcedirectio,hard,nointr,timeo=600,rsize=32768,wsize=32768,suid 0 0
Error i found, investigating this issue:
netapp.nfsd > hpuxia33: Flags [.], cksum 0xfa4c (correct), seq 0, ack 1, win 262, length 0
18:29:45.870431 IP (tos 0x0, ttl 64, id 24777, offset 0, flags [DF], proto TCP (6), length 61)
hpuxia33.1023 > netapp.nfsd: Flags [R], cksum 0x9d2d (correct), seq 1294045999:1294046020, win 0, length 21 [RST TCPS_SYN_SENT-Bad_seq]
18:29:46.154392 IP (tos 0x0, ttl 125, id 32480, offset 0, flags [DF], proto TCP (6), length 92)
We can see in that tcpdump, that RST (rest) signal issued by hpuxia node in SYN ACK handshake between servers because of BAD_seq (bad sequence) error.
This error fixed in PHNE_33159 HPUX patch for 11.11 OS, but i have 11.31 and i downloaded latest
cumulative ARPA transport patch PHNE_44266 for 11.31 but issue has not been resolved.
1. According to HPUX this issue fixed in latest NFS patch for 11.31:
Latest is B.18.104.22.168 (for today)
2. I had the same issue on 11.23 ITANIUM cluster. Latest NFS patch for it is: PHNE_43704
can be downloaded from here:
In my case issue was not related to HPUX.
3. To solve the problem next checks are useful:
a. Make sure that NFS client functionality is proper. Below tests can be done for this.
Create a test share from another node and export it via NFS. Then, try NFS mount of that test share in problematic server “hpuxia”. If its working fine, then, issue is not at NFS client end.
hpuxia2# mkdir /test (second hpux node)
hpuxia2# share -F nfs -o anon=2 /test
hpuxia# showmount –e hpuxia2 (first hpux node)
hpuxia# mkdir /test_nfs
hpuxia# mount -F nfs hpuxia2:/test /test_nfs
b. Verify the NFS product installed using “ # swverify –v ONCplus NFS ” . Review “/var/adm/sw/swagent.log” for any errors.
c. Step a will help us to confirm whether issue is at NFS client end or at NFS server end. If at client end, please try below.
Restart NFS services in below order
# /sbin/init.d/nfs.client stop
# /sbin/init.d/nfs.core stop
# /sbin/init.d/nfs.core start
# /sbin/init.d/nfs.client start
d. Test mount using lesser rsize and wsize
# mount –F nfs -o rsize=8192,wsize=8192 netapp:/ora_voting_disk/hpuxia /orarac
Most Users Ever Online: 31
Currently Browsing this Page:
Guest Posters: 0
Newest Members: Lo0oM
Administrators: Lo0oM (217)