Hi, I just upgraded from 6.2.2 to 6.4.2P6 and noticed that after creating a new qtree to be exported to a database client the 'exportfs -va' command produces a different behavior.
It appears that it unexports and then does a re-export. Before, on 6.2.2, it just did a re-export. Needles to say doing an unexport on a live mounted filesystem is not to nice.
Has anyone seen this as well ?
unexported /vol/auto2/its-adsprod759 re-exported /vol/auto2/its-adsprod759 unexported /vol/auto2/its-adsqa759 re-exported /vol/auto2/its-adsqa759 unexported /vol/auto2/its-hrtrn83 re-exported /vol/auto2/its-hrtrn83 unexported /vol/no2/cns-pbat-rec2 re-exported /vol/no2/cns-pbat-rec2 unexported /vol/no2/cns-pceo-rec2 re-exported /vol/no2/cns-pceo-rec2 unexported /vol/no2/cns-pdev-rec2 re-exported /vol/no2/cns-pdev-rec2 unexported /vol/no2/cns-pitsm-rec2 unexported /vol/dqa1/cns-test-dbf re-exported /vol/dqa1/cns-test-dbf unexported /vol/dqa1/its-ADSQA-dbf re-exported /vol/dqa1/its-ADSQA-dbf unexported /vol/dqa1/its-FIN70AU-dbf re-exported /vol/dqa1/its-FIN70AU-dbf
Thanks, George
------------------------------------------------------------------------------- George Kahler e-mail: george@yorku.ca Sr. Systems Administrator humans: (416) 736-2100 x.22699 Computing and Network Services machines: (416) 736-5830 Ontario, Canada, M3J-1P3
George,
It always did an unexport and then re-export. Your window of time is pretty small, it prepares the new export and then unexports the old one. There is no comparison of the nodes to determine if anything has changed in the rules.
I.e., it doesn't do a complete unexport of all nodes and then re-export everything.
In 6.5, instead of unexporting any path duplicates, the code will check to see if the node has actually changed and then only unexport and re-export.
As for doing it on a live mount, remember that NFS is stateless and well designed client applications need to be able to handle the occasional ESTALE. Or, this is the same as if the filer was rebooted, you expect the path to be re-exported and the client should not be able to tell the difference.
Slight humor aside, this is bug 101640 and was driven by the desire to not present ESTALE to database clients during the small window of time which the re-export is occuring on nodes which did not change.
Thanks, Tom
Hi, I just upgraded from 6.2.2 to 6.4.2P6 and noticed that after creating a new qtree to be exported to a database client the 'exportfs -va' command produces a different behavior.
It appears that it unexports and then does a re-export. Before, on 6.2.2, it just did a re-export. Needles to say doing an unexport on a live mounted filesystem is not to nice.
Has anyone seen this as well ?
unexported /vol/auto2/its-adsprod759 re-exported /vol/auto2/its-adsprod759 unexported /vol/auto2/its-adsqa759 re-exported /vol/auto2/its-adsqa759 unexported /vol/auto2/its-hrtrn83 re-exported /vol/auto2/its-hrtrn83 unexported /vol/no2/cns-pbat-rec2 re-exported /vol/no2/cns-pbat-rec2 unexported /vol/no2/cns-pceo-rec2 re-exported /vol/no2/cns-pceo-rec2 unexported /vol/no2/cns-pdev-rec2 re-exported /vol/no2/cns-pdev-rec2 unexported /vol/no2/cns-pitsm-rec2 unexported /vol/dqa1/cns-test-dbf re-exported /vol/dqa1/cns-test-dbf unexported /vol/dqa1/its-ADSQA-dbf re-exported /vol/dqa1/its-ADSQA-dbf unexported /vol/dqa1/its-FIN70AU-dbf re-exported /vol/dqa1/its-FIN70AU-dbf
Thanks, George
George Kahler e-mail: george@yorku.ca Sr. Systems Administrator humans: (416) 736-2100 x.22699 Computing and Network Services machines: (416) 736-5830 Ontario, Canada, M3J-1P3
OK, so wold the messages have changed then ? Because before when I did the 'exportfs -va' only the 're-exported' messages were echo(ed) back.
This is what used to come up on version 6.2.2
Fri Dec 5 16:22:53 EST [forrestal: rshd_0:auditlog]: :IN:rsh shell:RSH INPUT COMMAND is exportfs -va Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-adsprod759 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-adsqa759 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-hrtrn83 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-sasdata Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-home Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-sasdata Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-sasweb Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/sis-soft_kits Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-PSfin Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-PShr Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pbat-rec2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pceo-rec2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pdev-rec2
Thx. George
On Mon, 5 Jan 2004 13:29:31 -0800 (PST), "Haynes, Tom" thomas@netapp.com wrote:
George,
It always did an unexport and then re-export. Your window of time is pretty small, it prepares the new export and then unexports the old one. There is no comparison of the nodes to determine if anything has changed in the rules.
I.e., it doesn't do a complete unexport of all nodes and then re-export everything.
In 6.5, instead of unexporting any path duplicates, the code will check to see if the node has actually changed and then only unexport and re-export.
As for doing it on a live mount, remember that NFS is stateless and well designed client applications need to be able to handle the occasional ESTALE. Or, this is the same as if the filer was rebooted, you expect the path to be re-exported and the client should not be able to tell the difference.
Slight humor aside, this is bug 101640 and was driven by the desire to not present ESTALE to database clients during the small window of time which the re-export is occuring on nodes which did not change.
Thanks, Tom
Hi, I just upgraded from 6.2.2 to 6.4.2P6 and noticed that after creating a new qtree to be exported to a database client the 'exportfs -va' command produces a different behavior.
It appears that it unexports and then does a re-export. Before, on 6.2.2, it just did a re-export. Needles to say doing an unexport on a live mounted filesystem is not to nice.
Has anyone seen this as well ?
unexported /vol/auto2/its-adsprod759 re-exported /vol/auto2/its-adsprod759 unexported /vol/auto2/its-adsqa759 re-exported /vol/auto2/its-adsqa759 unexported /vol/auto2/its-hrtrn83 re-exported /vol/auto2/its-hrtrn83 unexported /vol/no2/cns-pbat-rec2 re-exported /vol/no2/cns-pbat-rec2 unexported /vol/no2/cns-pceo-rec2 re-exported /vol/no2/cns-pceo-rec2 unexported /vol/no2/cns-pdev-rec2 re-exported /vol/no2/cns-pdev-rec2 unexported /vol/no2/cns-pitsm-rec2 unexported /vol/dqa1/cns-test-dbf re-exported /vol/dqa1/cns-test-dbf unexported /vol/dqa1/its-ADSQA-dbf re-exported /vol/dqa1/its-ADSQA-dbf unexported /vol/dqa1/its-FIN70AU-dbf re-exported /vol/dqa1/its-FIN70AU-dbf
Thanks, George
George Kahler e-mail: george@yorku.ca Sr. Systems Administrator humans: (416) 736-2100 x.22699 Computing and Network Services machines: (416) 736-5830 Ontario, Canada, M3J-1P3
-- Tom Haynes, ex-cfb thomas@netapp.com
OK, so wold the messages have changed then ?
Yes, the exports code has been steadily improved over the last couple of years. ;>
Because before when I did the 'exportfs -va' only the 're-exported' messages were echo(ed) back.
There probabaly were several areas of code which did un-exporting and not all of them were logging the fact. When I did the consolidation, a new feature was that the re-exporting started logging the un-export.
Check to see if this was the set of all hosts at the time of the exportfs -a.
This is what used to come up on version 6.2.2
Fri Dec 5 16:22:53 EST [forrestal: rshd_0:auditlog]: :IN:rsh shell:RSH INPUT COMMAND is exportfs -va Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-adsprod759 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-adsqa759 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-hrtrn83 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-sasdata Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-home Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-sasdata Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-sasweb Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/sis-soft_kits Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-PSfin Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-PShr Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pbat-rec2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pceo-rec2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pdev-rec2
Thx. George
On Mon, 5 Jan 2004 13:29:31 -0800 (PST), "Haynes, Tom" thomas@netapp.com wrote:
George,
It always did an unexport and then re-export. Your window of time is pretty small, it prepares the new export and then unexports the old one. There is no comparison of the nodes to determine if anything has changed in the rules.
I.e., it doesn't do a complete unexport of all nodes and then re-export everything.
In 6.5, instead of unexporting any path duplicates, the code will check to see if the node has actually changed and then only unexport and re-export.
As for doing it on a live mount, remember that NFS is stateless and well designed client applications need to be able to handle the occasional ESTALE. Or, this is the same as if the filer was rebooted, you expect the path to be re-exported and the client should not be able to tell the difference.
Slight humor aside, this is bug 101640 and was driven by the desire to not present ESTALE to database clients during the small window of time which the re-export is occuring on nodes which did not change.
Thanks, Tom
Hi, I just upgraded from 6.2.2 to 6.4.2P6 and noticed that after creating a new qtree to be exported to a database client the 'exportfs -va' command produces a different behavior.
It appears that it unexports and then does a re-export. Before, on 6.2.2, it just did a re-export. Needles to say doing an unexport on a live mounted filesystem is not to nice.
Has anyone seen this as well ?
unexported /vol/auto2/its-adsprod759 re-exported /vol/auto2/its-adsprod759 unexported /vol/auto2/its-adsqa759 re-exported /vol/auto2/its-adsqa759 unexported /vol/auto2/its-hrtrn83 re-exported /vol/auto2/its-hrtrn83 unexported /vol/no2/cns-pbat-rec2 re-exported /vol/no2/cns-pbat-rec2 unexported /vol/no2/cns-pceo-rec2 re-exported /vol/no2/cns-pceo-rec2 unexported /vol/no2/cns-pdev-rec2 re-exported /vol/no2/cns-pdev-rec2 unexported /vol/no2/cns-pitsm-rec2 unexported /vol/dqa1/cns-test-dbf re-exported /vol/dqa1/cns-test-dbf unexported /vol/dqa1/its-ADSQA-dbf re-exported /vol/dqa1/its-ADSQA-dbf unexported /vol/dqa1/its-FIN70AU-dbf re-exported /vol/dqa1/its-FIN70AU-dbf
Thanks, George
George Kahler e-mail: george@yorku.ca Sr. Systems Administrator humans: (416) 736-2100 x.22699 Computing and Network Services machines: (416) 736-5830 Ontario, Canada, M3J-1P3
-- Tom Haynes, ex-cfb thomas@netapp.com
Tom, thanks for the explanation ! George
On Mon, 5 Jan 2004 13:49:19 -0800 (PST), "Haynes, Tom" thomas@netapp.com wrote:
OK, so wold the messages have changed then ?
Yes, the exports code has been steadily improved over the last couple of years. ;>
Because before when I did the 'exportfs -va' only the 're-exported' messages were echo(ed) back.
There probabaly were several areas of code which did un-exporting and not all of them were logging the fact. When I did the consolidation, a new feature was that the re-exporting started logging the un-export.
Check to see if this was the set of all hosts at the time of the exportfs -a.
This is what used to come up on version 6.2.2
Fri Dec 5 16:22:53 EST [forrestal: rshd_0:auditlog]: :IN:rsh shell:RSH INPUT COMMAND is exportfs -va Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-adsprod759 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-adsqa759 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-hrtrn83 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-sasdata Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-home Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-sasdata Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/mi-sasweb Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/sis-soft_kits Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-PSfin Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/auto2/its-PShr Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pbat-rec2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pceo-rec2 Fri Dec 5 16:22:54 EST [forrestal: rshd_0:auditlog]: root:OUT:exportfs:re-exported /vol/no2/cns-pdev-rec2
Thx. George
On Mon, 5 Jan 2004 13:29:31 -0800 (PST), "Haynes, Tom" thomas@netapp.com wrote:
George,
It always did an unexport and then re-export. Your window of time is pretty small, it prepares the new export and then unexports the old one. There is no comparison of the nodes to determine if anything has changed in the rules.
I.e., it doesn't do a complete unexport of all nodes and then re-export everything.
In 6.5, instead of unexporting any path duplicates, the code will check to see if the node has actually changed and then only unexport and re-export.
As for doing it on a live mount, remember that NFS is stateless and well designed client applications need to be able to handle the occasional ESTALE. Or, this is the same as if the filer was rebooted, you expect the path to be re-exported and the client should not be able to tell the difference.
Slight humor aside, this is bug 101640 and was driven by the desire to not present ESTALE to database clients during the small window of time which the re-export is occuring on nodes which did not change.
Thanks, Tom
Hi, I just upgraded from 6.2.2 to 6.4.2P6 and noticed that after creating a new qtree to be exported to a database client the 'exportfs -va' command produces a different behavior.
It appears that it unexports and then does a re-export. Before, on 6.2.2, it just did a re-export. Needles to say doing an unexport on a live mounted filesystem is not to nice.
Has anyone seen this as well ?
unexported /vol/auto2/its-adsprod759 re-exported /vol/auto2/its-adsprod759 unexported /vol/auto2/its-adsqa759 re-exported /vol/auto2/its-adsqa759 unexported /vol/auto2/its-hrtrn83 re-exported /vol/auto2/its-hrtrn83 unexported /vol/no2/cns-pbat-rec2 re-exported /vol/no2/cns-pbat-rec2 unexported /vol/no2/cns-pceo-rec2 re-exported /vol/no2/cns-pceo-rec2 unexported /vol/no2/cns-pdev-rec2 re-exported /vol/no2/cns-pdev-rec2 unexported /vol/no2/cns-pitsm-rec2 unexported /vol/dqa1/cns-test-dbf re-exported /vol/dqa1/cns-test-dbf unexported /vol/dqa1/its-ADSQA-dbf re-exported /vol/dqa1/its-ADSQA-dbf unexported /vol/dqa1/its-FIN70AU-dbf re-exported /vol/dqa1/its-FIN70AU-dbf
Thanks, George
George Kahler e-mail: george@yorku.ca Sr. Systems Administrator humans: (416) 736-2100 x.22699 Computing and Network Services machines: (416) 736-5830 Ontario, Canada, M3J-1P3
-- Tom Haynes, ex-cfb thomas@netapp.com
-- Tom Haynes, ex-cfb thomas@netapp.com
-
George Kahler -
Haynes, Tom