Thank you Justin for your response.
Since this is a four-node cluster, if I force this application’s CIFS traffic through a lif on HA-pair “A” to a volume on an aggr owned by a controller in HA-pair “B” would those sessions still be severed if I upgrade HA-pair “B”? Would “B”s takeover/giveback process of changing aggr ownership also require severing CIFS sessions?
On Mar 13, 2019, at 8:50 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
Microsoft and NetApp stance is that CA shares are supported only for HyperV and SQL workloads at this time.
-----Original Message----- From: toasters-bounces@teaparty.net toasters-bounces@teaparty.net On Behalf Of s.eno via Toasters Sent: Wednesday, March 13, 2019 8:39 AM To: toasters@teaparty.net Subject:
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
On Mar 13, 2019, at 8:39 AM, s.eno s.eno@me.com wrote:
Hi,
Does anyone here utilize continuously available CIFS shares for anything other than Hyper-V and/or SQL? If so, do you have real-world results on how those shares do during ONTAP upgrades?
I need to perform upgrades on a 4-node cluster and have a mission critical application that doesn’t like it when the takeover/giveback process severs its CIFS sessions. I’m trying to figure out a way to do this without causing downtime to the application which in-turn causes push-back on the upgrade plan.
The application appears to be connecting to the shares via SMB3_1, so it should support CA, right? I’ve tested checking the CA box on one of its shares and the app drains connections to the share and stops sending any more traffic, so it appears to not like something related to that change. One of its shares, however, was provisioned with that box checked and that one works just fine as far as traffic & sessions go.
-- Scott s.eno@me.com
The sessions will live on the node where the connection is. That node's nblade keeps the locks, which is the source of the blips that happen on takeovers. So you should be fine.
SMB 2.x and 3.x are pretty resilient, so the blip is often minor.
-----Original Message----- From: SE s.eno@me.com Sent: Wednesday, March 13, 2019 9:06 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you Justin for your response.
Since this is a four-node cluster, if I force this application’s CIFS traffic through a lif on HA-pair “A” to a volume on an aggr owned by a controller in HA-pair “B” would those sessions still be severed if I upgrade HA-pair “B”? Would “B”s takeover/giveback process of changing aggr ownership also require severing CIFS sessions?
On Mar 13, 2019, at 8:50 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
Microsoft and NetApp stance is that CA shares are supported only for HyperV and SQL workloads at this time.
-----Original Message----- From: toasters-bounces@teaparty.net toasters-bounces@teaparty.net On Behalf Of s.eno via Toasters Sent: Wednesday, March 13, 2019 8:39 AM To: toasters@teaparty.net Subject:
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
On Mar 13, 2019, at 8:39 AM, s.eno s.eno@me.com wrote:
Hi,
Does anyone here utilize continuously available CIFS shares for anything other than Hyper-V and/or SQL? If so, do you have real-world results on how those shares do during ONTAP upgrades?
I need to perform upgrades on a 4-node cluster and have a mission critical application that doesn’t like it when the takeover/giveback process severs its CIFS sessions. I’m trying to figure out a way to do this without causing downtime to the application which in-turn causes push-back on the upgrade plan.
The application appears to be connecting to the shares via SMB3_1, so it should support CA, right? I’ve tested checking the CA box on one of its shares and the app drains connections to the share and stops sending any more traffic, so it appears to not like something related to that change. One of its shares, however, was provisioned with that box checked and that one works just fine as far as traffic & sessions go.
-- Scott s.eno@me.com
Thank you very much for the assistance Justin.
It is frustrating because as you say the application should recover, especially if it is connecting via SMB3_1, but all previous upgrades showed this application not recovering and forcing reboots of its application servers to get it functional again.
-- Scott s.eno@me.com
On Mar 13, 2019, at 9:36 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
The sessions will live on the node where the connection is. That node's nblade keeps the locks, which is the source of the blips that happen on takeovers. So you should be fine.
SMB 2.x and 3.x are pretty resilient, so the blip is often minor.
-----Original Message----- From: SE s.eno@me.com Sent: Wednesday, March 13, 2019 9:06 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you Justin for your response.
Since this is a four-node cluster, if I force this application’s CIFS traffic through a lif on HA-pair “A” to a volume on an aggr owned by a controller in HA-pair “B” would those sessions still be severed if I upgrade HA-pair “B”? Would “B”s takeover/giveback process of changing aggr ownership also require severing CIFS sessions?
On Mar 13, 2019, at 8:50 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
Microsoft and NetApp stance is that CA shares are supported only for HyperV and SQL workloads at this time.
-----Original Message----- From: toasters-bounces@teaparty.net toasters-bounces@teaparty.net On Behalf Of s.eno via Toasters Sent: Wednesday, March 13, 2019 8:39 AM To: toasters@teaparty.net Subject:
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
On Mar 13, 2019, at 8:39 AM, s.eno s.eno@me.com wrote:
Hi,
Does anyone here utilize continuously available CIFS shares for anything other than Hyper-V and/or SQL? If so, do you have real-world results on how those shares do during ONTAP upgrades?
I need to perform upgrades on a 4-node cluster and have a mission critical application that doesn’t like it when the takeover/giveback process severs its CIFS sessions. I’m trying to figure out a way to do this without causing downtime to the application which in-turn causes push-back on the upgrade plan.
The application appears to be connecting to the shares via SMB3_1, so it should support CA, right? I’ve tested checking the CA box on one of its shares and the app drains connections to the share and stops sending any more traffic, so it appears to not like something related to that change. One of its shares, however, was provisioned with that box checked and that one works just fine as far as traffic & sessions go.
-- Scott s.eno@me.com
Hmm. Have you engaged the application vendor to see if there is some sort of configuration option you can toggle?
Or perhaps root cause why the upgrades are causing the issue?
-----Original Message----- From: s.eno s.eno@me.com Sent: Wednesday, March 13, 2019 10:04 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you very much for the assistance Justin.
It is frustrating because as you say the application should recover, especially if it is connecting via SMB3_1, but all previous upgrades showed this application not recovering and forcing reboots of its application servers to get it functional again.
-- Scott s.eno@me.com
On Mar 13, 2019, at 9:36 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
The sessions will live on the node where the connection is. That node's nblade keeps the locks, which is the source of the blips that happen on takeovers. So you should be fine.
SMB 2.x and 3.x are pretty resilient, so the blip is often minor.
-----Original Message----- From: SE s.eno@me.com Sent: Wednesday, March 13, 2019 9:06 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you Justin for your response.
Since this is a four-node cluster, if I force this application’s CIFS traffic through a lif on HA-pair “A” to a volume on an aggr owned by a controller in HA-pair “B” would those sessions still be severed if I upgrade HA-pair “B”? Would “B”s takeover/giveback process of changing aggr ownership also require severing CIFS sessions?
On Mar 13, 2019, at 8:50 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
Microsoft and NetApp stance is that CA shares are supported only for HyperV and SQL workloads at this time.
-----Original Message----- From: toasters-bounces@teaparty.net toasters-bounces@teaparty.net On Behalf Of s.eno via Toasters Sent: Wednesday, March 13, 2019 8:39 AM To: toasters@teaparty.net Subject:
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
On Mar 13, 2019, at 8:39 AM, s.eno s.eno@me.com wrote:
Hi,
Does anyone here utilize continuously available CIFS shares for anything other than Hyper-V and/or SQL? If so, do you have real-world results on how those shares do during ONTAP upgrades?
I need to perform upgrades on a 4-node cluster and have a mission critical application that doesn’t like it when the takeover/giveback process severs its CIFS sessions. I’m trying to figure out a way to do this without causing downtime to the application which in-turn causes push-back on the upgrade plan.
The application appears to be connecting to the shares via SMB3_1, so it should support CA, right? I’ve tested checking the CA box on one of its shares and the app drains connections to the share and stops sending any more traffic, so it appears to not like something related to that change. One of its shares, however, was provisioned with that box checked and that one works just fine as far as traffic & sessions go.
-- Scott s.eno@me.com
Another idea: since it's a SMB 3 connection, how about *multipathing*? Supported since 9.4 on ONTAP... Appeared in System Manager in 9.5.
Have 2 LIFs on the usual node. Move them one after the other nondisruptively to the other HA pair before an upgrade and back, when you're done.
Should be completely non-disruptive to the connection/session, or what do you think, Justin?
Greetings from Northfriesland
Sebastian
On Wed, Mar 13, 2019, 15:22 Parisi, Justin Justin.Parisi@netapp.com wrote:
Hmm. Have you engaged the application vendor to see if there is some sort of configuration option you can toggle?
Or perhaps root cause why the upgrades are causing the issue?
-----Original Message----- From: s.eno s.eno@me.com Sent: Wednesday, March 13, 2019 10:04 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you very much for the assistance Justin.
It is frustrating because as you say the application should recover, especially if it is connecting via SMB3_1, but all previous upgrades showed this application not recovering and forcing reboots of its application servers to get it functional again.
-- Scott s.eno@me.com
On Mar 13, 2019, at 9:36 AM, Parisi, Justin Justin.Parisi@netapp.com
wrote:
The sessions will live on the node where the connection is. That node's
nblade keeps the locks, which is the source of the blips that happen on takeovers. So you should be fine.
SMB 2.x and 3.x are pretty resilient, so the blip is often minor.
-----Original Message----- From: SE s.eno@me.com Sent: Wednesday, March 13, 2019 9:06 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links
or open attachments unless you recognize the sender and know the content is safe.
Thank you Justin for your response.
Since this is a four-node cluster, if I force this application’s CIFS
traffic through a lif on HA-pair “A” to a volume on an aggr owned by a controller in HA-pair “B” would those sessions still be severed if I upgrade HA-pair “B”? Would “B”s takeover/giveback process of changing aggr ownership also require severing CIFS sessions?
On Mar 13, 2019, at 8:50 AM, Parisi, Justin Justin.Parisi@netapp.com
wrote:
Microsoft and NetApp stance is that CA shares are supported only for
HyperV and SQL workloads at this time.
-----Original Message----- From: toasters-bounces@teaparty.net toasters-bounces@teaparty.net On
Behalf Of s.eno via Toasters
Sent: Wednesday, March 13, 2019 8:39 AM To: toasters@teaparty.net Subject:
NetApp Security WARNING: This is an external email. Do not click links
or open attachments unless you recognize the sender and know the content is safe.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
On Mar 13, 2019, at 8:39 AM, s.eno s.eno@me.com wrote:
Hi,
Does anyone here utilize continuously available CIFS shares for anything
other than Hyper-V and/or SQL? If so, do you have real-world results on how those shares do during ONTAP upgrades?
I need to perform upgrades on a 4-node cluster and have a mission
critical application that doesn’t like it when the takeover/giveback process severs its CIFS sessions. I’m trying to figure out a way to do this without causing downtime to the application which in-turn causes push-back on the upgrade plan.
The application appears to be connecting to the shares via SMB3_1, so it
should support CA, right? I’ve tested checking the CA box on one of its shares and the app drains connections to the share and stops sending any more traffic, so it appears to not like something related to that change. One of its shares, however, was provisioned with that box checked and that one works just fine as far as traffic & sessions go.
-- Scott s.eno@me.com
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
I am definitely looking forward to this capability, but we’re at 9.1P2 now. I’m trying to get to 9.5 eventually, but have to stop at 9.3P11 first. But yes, I’ve read about SMB multipathing and am anxious to get it implemented for the reason of my original post.
-- Scott s.eno@me.com
On Mar 28, 2019, at 2:23 AM, Sebastian P. Goetze spgoetze@gmail.com wrote:
Another idea: since it's a SMB 3 connection, how about *multipathing*? Supported since 9.4 on ONTAP... Appeared in System Manager in 9.5.
Have 2 LIFs on the usual node. Move them one after the other nondisruptively to the other HA pair before an upgrade and back, when you're done.
Should be completely non-disruptive to the connection/session, or what do you think, Justin?
Greetings from Northfriesland
Sebastian
On Wed, Mar 13, 2019, 15:22 Parisi, Justin Justin.Parisi@netapp.com wrote: Hmm. Have you engaged the application vendor to see if there is some sort of configuration option you can toggle?
Or perhaps root cause why the upgrades are causing the issue?
-----Original Message----- From: s.eno s.eno@me.com Sent: Wednesday, March 13, 2019 10:04 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you very much for the assistance Justin.
It is frustrating because as you say the application should recover, especially if it is connecting via SMB3_1, but all previous upgrades showed this application not recovering and forcing reboots of its application servers to get it functional again.
-- Scott s.eno@me.com
On Mar 13, 2019, at 9:36 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
The sessions will live on the node where the connection is. That node's nblade keeps the locks, which is the source of the blips that happen on takeovers. So you should be fine.
SMB 2.x and 3.x are pretty resilient, so the blip is often minor.
-----Original Message----- From: SE s.eno@me.com Sent: Wednesday, March 13, 2019 9:06 AM To: Parisi, Justin Justin.Parisi@netapp.com Cc: Toasters toasters@teaparty.net Subject: Re: Continuously available CIFS
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Thank you Justin for your response.
Since this is a four-node cluster, if I force this application’s CIFS traffic through a lif on HA-pair “A” to a volume on an aggr owned by a controller in HA-pair “B” would those sessions still be severed if I upgrade HA-pair “B”? Would “B”s takeover/giveback process of changing aggr ownership also require severing CIFS sessions?
On Mar 13, 2019, at 8:50 AM, Parisi, Justin Justin.Parisi@netapp.com wrote:
Microsoft and NetApp stance is that CA shares are supported only for HyperV and SQL workloads at this time.
-----Original Message----- From: toasters-bounces@teaparty.net toasters-bounces@teaparty.net On Behalf Of s.eno via Toasters Sent: Wednesday, March 13, 2019 8:39 AM To: toasters@teaparty.net Subject:
NetApp Security WARNING: This is an external email. Do not click links or open attachments unless you recognize the sender and know the content is safe.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
On Mar 13, 2019, at 8:39 AM, s.eno s.eno@me.com wrote:
Hi,
Does anyone here utilize continuously available CIFS shares for anything other than Hyper-V and/or SQL? If so, do you have real-world results on how those shares do during ONTAP upgrades?
I need to perform upgrades on a 4-node cluster and have a mission critical application that doesn’t like it when the takeover/giveback process severs its CIFS sessions. I’m trying to figure out a way to do this without causing downtime to the application which in-turn causes push-back on the upgrade plan.
The application appears to be connecting to the shares via SMB3_1, so it should support CA, right? I’ve tested checking the CA box on one of its shares and the app drains connections to the share and stops sending any more traffic, so it appears to not like something related to that change. One of its shares, however, was provisioned with that box checked and that one works just fine as far as traffic & sessions go.
-- Scott s.eno@me.com
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
-- sent from my mobile, spellcheck might have messed up...
-
Parisi, Justin -
s.eno -
SE
-
Sebastian P. Goetze