September 2019 - Devel - NFS Ganesha List Archives

Change in ...nfs-ganesha[next]: WIP: FSAL_GLUSTER: Improving fd handling during OPEN and COMMIT
by Soumya (GerritHub) 19 Sep '19

19 Sep '19

Soumya has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468806 ) Change subject: WIP: FSAL_GLUSTER: Improving fd handling during OPEN and COMMIT ...................................................................... WIP: FSAL_GLUSTER: Improving fd handling during OPEN and COMMIT a) At present, in FSAL_GLUSTER, while opening a file the very first time, we use two APIs * glfs_h_creat to create (file &)handle * glfs_h_open to fetch a file descriptor This results in opening the file twice (once during create and another during open) which is inefficient and also may result in EACCESS while creating a read-only file. To address that, have added a gluster API "glfs_h_creat_open" which creates and returns both handle and glfd. - https://review.gluster.org/#/c/glusterfs/+/23448 This API can be used to atomically fetch both handle and glfd, thus improving performance. b) In addition, for stateless operations like COMMIT etc, FSAL tries to open a new fd. This again may result in EACCESS errors while operating on read-only files. More details are captured in https://bugzilla.redhat.com/show_bug.cgi?id=1735480#c1 To address this issue, we can save the fd created during open2 as globalfd, thus improving performance and also providing correctness as expected by POSIX. Change-Id: I55f734875e9d803148d2a19bf090f04d537f9ebf Signed-off-by: Soumya Koduri <skoduri(a)redhat.com> --- M src/FSAL/FSAL_GLUSTER/handle.c 1 file changed, 178 insertions(+), 25 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/06/468806/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468806 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: I55f734875e9d803148d2a19bf090f04d537f9ebf Gerrit-Change-Number: 468806 Gerrit-PatchSet: 1 Gerrit-Owner: Soumya <skoduri(a)redhat.com> Gerrit-MessageType: newchange

1 0

Change in ...nfs-ganesha[next]: Reopen the file only if openflags change
by Soumya (GerritHub) 19 Sep '19

19 Sep '19

Soumya has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468805 ) Change subject: Reopen the file only if openflags change ...................................................................... Reopen the file only if openflags change If a file is already opened with required openflags, no need to reopen it. Change-Id: I475a18c4692810e7d3851e07a1c05030a16f7975 Signed-off-by: Soumya Koduri <skoduri(a)redhat.com> --- M src/FSAL/FSAL_CEPH/handle.c M src/FSAL/FSAL_GLUSTER/handle.c M src/FSAL/FSAL_GPFS/file.c M src/FSAL/FSAL_LIZARDFS/handle.c M src/FSAL/FSAL_MEM/mem_handle.c M src/FSAL/FSAL_RGW/handle.c M src/FSAL/FSAL_VFS/file.c 7 files changed, 44 insertions(+), 2 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/05/468805/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468805 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: I475a18c4692810e7d3851e07a1c05030a16f7975 Gerrit-Change-Number: 468805 Gerrit-PatchSet: 1 Gerrit-Owner: Soumya <skoduri(a)redhat.com> Gerrit-MessageType: newchange

1 0

Change in ...nfs-ganesha[next]: Handle ETIMEDOUT error from DMAPI environments
by Malahal (GerritHub) 18 Sep '19

18 Sep '19

Malahal has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468764 ) Change subject: Handle ETIMEDOUT error from DMAPI environments ...................................................................... Handle ETIMEDOUT error from DMAPI environments ETIMEDOUT is returned from some file systems that recall files from tapes. Handle these with NFS3ERR_JUKEBOX NFS4ERR_DELAY so that NFS clients retry later rather than failing applications. Change-Id: Ic5b045f7e8d18c9ab60108253299f7ef54e83c6a Signed-off-by: Malahal Naineni <malahal(a)us.ibm.com> --- M src/FSAL/fsal_convert.c 1 file changed, 1 insertion(+), 0 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/64/468764/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468764 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: Ic5b045f7e8d18c9ab60108253299f7ef54e83c6a Gerrit-Change-Number: 468764 Gerrit-PatchSet: 1 Gerrit-Owner: Malahal <malahal(a)gmail.com> Gerrit-MessageType: newchange

1 0

Change in ...nfs-ganesha[next]: MDCACHE - do not print CRIT message when fsal_close() fails.
by Name of user not set (GerritHub) 18 Sep '19

18 Sep '19

pradeepthomas(a)gmail.com has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468757 ) Change subject: MDCACHE - do not print CRIT message when fsal_close() fails. ...................................................................... MDCACHE - do not print CRIT message when fsal_close() fails. It is possible that the file is not opened by the FSAL at the time of cleanup. So ignore the FSAL_NOT_OPENED error. Change-Id: I014dde1b5f60cbd1877c84e5f7d97bfe5c32108a Signed-off-by: Pradeep <pradeepthomas(a)gmail.com> --- M src/FSAL/Stackable_FSALs/FSAL_MDCACHE/mdcache_lru.c 1 file changed, 2 insertions(+), 1 deletion(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/57/468757/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468757 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: I014dde1b5f60cbd1877c84e5f7d97bfe5c32108a Gerrit-Change-Number: 468757 Gerrit-PatchSet: 1 Gerrit-Owner: pradeepthomas(a)gmail.com Gerrit-MessageType: newchange

1 0

Change in ...nfs-ganesha[next]: heap use after free for component_log_level
by Suhrud Patankar (GerritHub) 18 Sep '19

18 Sep '19

Suhrud Patankar has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468729 ) Change subject: heap use after free for component_log_level ...................................................................... heap use after free for component_log_level In function log_conf_commit: 2568 if (component_log_level != default_log_levels) 2569 gsh_free(component_log_level); 2570 component_log_level = logger->comp_log_level; component_log_level is a global pointer. While the current thread is freeing and re-setting, If any other thread is accessing the pointer after execution of line 2569 but before 2570 is fully executed it will hit use-after-free case. Avoid changing component_log_level once set. Change-Id: If44236e323f39473c590930f4ce8bd4fd127c5f7 Signed-off-by: ashish.sangwan <ashish.sangwan(a)nutanix.com> --- M src/log/log_functions.c 1 file changed, 12 insertions(+), 5 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/29/468729/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468729 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: If44236e323f39473c590930f4ce8bd4fd127c5f7 Gerrit-Change-Number: 468729 Gerrit-PatchSet: 1 Gerrit-Owner: Suhrud Patankar <suhrudpatankar(a)gmail.com> Gerrit-MessageType: newchange

1 0

Change in ...nfs-ganesha[next]: build: add dbus_status_reply() to exports from libganesha_nfsd.so
by Kaleb KEITHLEY (GerritHub) 18 Sep '19

18 Sep '19

Kaleb KEITHLEY has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468728 ) Change subject: build: add dbus_status_reply() to exports from libganesha_nfsd.so ...................................................................... build: add dbus_status_reply() to exports from libganesha_nfsd.so An out-of-tree FSAL (FSAL_ATAFS ?) calls dbus_status_reply(). Signed-off-by: Kaleb S. KEITHLEY <kkeithle(a)redhat.com> Change-Id: Ib8fcf2a498321f64312a372c0ea55d91666d4f37 --- M src/MainNFSD/libganesha_nfsd.ver 1 file changed, 1 insertion(+), 0 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/28/468728/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468728 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: Ib8fcf2a498321f64312a372c0ea55d91666d4f37 Gerrit-Change-Number: 468728 Gerrit-PatchSet: 1 Gerrit-Owner: Kaleb KEITHLEY <kaleb(a)redhat.com> Gerrit-MessageType: newchange

1 0

Change in ...nfs-ganesha[next]: Add expiry to the client specific grace periods for recovery
by Sriram Patil (GerritHub) 15 Sep '19

15 Sep '19

Sriram Patil has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468406 ) Change subject: Add expiry to the client specific grace periods for recovery ...................................................................... Add expiry to the client specific grace periods for recovery The client specific grace periods are controlled by cid_reclaim_complete for NFSv4.1. The server stops the grace for that client when it sends a reclaim_complete request. However, if for some reason there is no reclaim_complete received from the client, the client is stuck while opening a file forever because ganesha keeps returning NFS4ERR_GRACE. Adding a timeout to make sure we do not end up in the hung state because of NFS client bug. THis fix prioritizes failing a reclaim request over waiting for a reclaim_complete indefinitely. Change-Id: Ia1ab1a49bff0d053c78b7606f723cb3fe82c80fb Signed-off-by: Sriram Patil <sriramp(a)vmware.com> --- M src/Protocols/NFS/nfs4_op_open.c M src/SAL/nfs4_clientid.c M src/SAL/nfs4_recovery.c M src/include/sal_data.h M src/include/sal_functions.h 5 files changed, 67 insertions(+), 3 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/06/468406/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468406 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: Ia1ab1a49bff0d053c78b7606f723cb3fe82c80fb Gerrit-Change-Number: 468406 Gerrit-PatchSet: 1 Gerrit-Owner: Sriram Patil <sriramp(a)vmware.com> Gerrit-MessageType: newchange

1 0

PROGRAM_NOT_AVAILABLE at server boot up causing IO error
by Suhrud Patankar 15 Sep '19

15 Sep '19

Hello All, At server boot, Ganesha registers with rpcbind only after loading all the exports. If export loading takes some time, then there is a window, where clients can get PROGRAM_NOT_AVAILABLE error. What I see is, the client tries to get NLM port number only a few times before converting to IO error. #~/multilock/ml_posix_client OPEN 0 RW create foo 1 OPEN OK 0 3 #### Reboot the server after this.##### LOCKW 0 read 10 10 2 LOCKW ERRNO 5 "Input/output error" "Lock failed" bad token "read 10 10" I understand loading the exports fast will reduce this window but the window will still be there. Also FSALs can take longer to create exports if there are high number of exports. Has anyone seen this? Is it possible to register with rpcbind before loading the exports? Thanks & Regards, Suhrud

4 3

Announce Push of V2.9-dev.8
by Frank Filz 13 Sep '19

13 Sep '19

Branch next Tag:V2.9-dev.8 NOTE: This merge has an ntirpc pullup, please update your submodule Release Highlights * Fix max open files limit for ganesha daemon * Remove inline comments from systemd unit files * Pass null-terminated string to strstr() * Add rate limited logging. * PROXY : separate export session by exportid * FSAL_RGW - On open/create, use new handle not parent * Delete SHARE state after the lock stats * Add LTTng traces to nfs_rpc set * Use xdr_putbufs for READ4resok annd READ3resok ntirpc pullup includes: * xdr_putbufs * nonblocking sendmsg for sending responses * vectorized RPCSEC_GSS Signed-off-by: Frank S. Filz <ffilzlnx(a)mindspring.com> Contents: 3c0e2de Frank S. Filz Use xdr_putbufs for READ4resok annd READ3resok 29cfd43 Frank S. Filz Add LTTng traces to nfs_rpc set d805ed3 Suhrud Patankar Delete SHARE state after the lock stats 7dd27e9 Daniel Gryniewicz FSAL_RGW - On open/create, use new handle not parent fbe4217 Patrice LUCAS PROXY : separate export session by exportid f5bb2ed Malahal Naineni Add rate limited logging. 7ffb8a9 Malahal Naineni Pass null-terminated string to strstr() 499c9c5 Malahal Naineni Remove inline comments from systemd unit files 6c74867 Malahal Naineni Fix max open files limit for ganesha daemon

1 0

Change in ...nfs-ganesha[next]: python: run setup.py for gpfs-epoch(.py) to fix its shebang
by Kaleb KEITHLEY (GerritHub) 12 Sep '19

12 Sep '19

Kaleb KEITHLEY has uploaded this change for review. ( https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468272 ) Change subject: python: run setup.py for gpfs-epoch(.py) to fix its shebang ...................................................................... python: run setup.py for gpfs-epoch(.py) to fix its shebang `setup.py build ...` modifies the shebang appropriately for the platform, i.e. using the python (e.g. /usr/bin/python3) that setup.py was invoked with. This might seem like a heavyweight solution for one file, but it's ostensibly better than some other solution like sed editing the shebang in an rpmbuild scriptlet. This way you get (almost) the right thing in a regular cmake/make/make-install build. Almost, because the `setup.py install ...` step installs gpfs-epoch in /usr/bin instead of where we infer it's wanted, in /usr/libexec, based on the install scriptlets in the rpm .spec file. I contemplated changing the --prefix for the `setup.py install ...` step but then that scribbles other things in /usr/libexec or /usr/local/libexec I also contemplated putting gpfs-epoch(.py) in its own subdir, e.g., .../src/scripts/gpfs-epoch/... or .../src/FSAL/FSAL_GPFS/gpfs-epoch/... and could be convinced to do that. If I did that than I could set the install prefix to /usr/libexec without breaking other things. Note: All versions of SUSE that we support (i.e. that we build community packages for in OBS) have python3 available. Signed-off-by: Kaleb S. KEITHLEY <kkeithle(a)redhat.com> Change-Id: Iaa328e3767597054fd0b0fac8b4c635f751bde46 --- M src/nfs-ganesha.spec-in.cmake M src/scripts/CMakeLists.txt M src/scripts/ganeshactl/CMakeLists.txt R src/scripts/gpfs-epoch.py A src/scripts/setup.py.in 5 files changed, 79 insertions(+), 2 deletions(-) git pull ssh://review.gerrithub.io:29418/ffilz/nfs-ganesha refs/changes/72/468272/1 -- To view, visit https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/468272 To unsubscribe, or for help writing mail filters, visit https://review.gerrithub.io/settings Gerrit-Project: ffilz/nfs-ganesha Gerrit-Branch: next Gerrit-Change-Id: Iaa328e3767597054fd0b0fac8b4c635f751bde46 Gerrit-Change-Number: 468272 Gerrit-PatchSet: 1 Gerrit-Owner: Kaleb KEITHLEY <kaleb(a)redhat.com> Gerrit-MessageType: newchange

1 0