summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
* nfs.conf: Remove static variables in parsing routinesJustin Mitchell2017-06-061-18/+21
| | | | | | | | | | | Part of a sequence of attempts to tidy up the nfs.conf code and prepare it for use as part of a configuration API. Remove static vars that prevented memory cleanup and potentially lead to parsing errors if conf_init was called again. Signed-off-by: Justin Mitchell <jumitche@rehat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: added 'u' to argument list.Steve Dickson2017-06-011-1/+1
| | | | | Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1451568 Signed-off-by: Steve Dickson <steved@redhat.com>
* mount.nfs: improve version negotiation when vers=4 is specified.NeilBrown2017-06-011-5/+9
| | | | | | | | | | | | | | | | | | | | | | | | If NFSv4, in general, is requested (possibly by -t nfs4 or -o v4 or -o vers=4 etc) then we need to negotiate the best minor version, but must not fallback to v3 or v2. Internally, this state is reflected in v_mode == V_GENERAL. This means that a major version was given, but the minor version still needs to be negotiated. This is handled by nfs_autonegotiate(). It currently does the right thing for EPROTONOSUPPORT and EINVAL, but not for other errors. In particular, ENOENT can cause problems as NFSv4 might export a different namespace than NFSv3 (e.g. by using fsid=0 in the Linux NFS server). Currently a mount request for NFSv4 and a particular path can result if an NFSv3 mount if the path is available with v3 but not v4. So move the special handling of V_GENERAL into the common fall_back: code, and add extra checking in the ENCONNREFUSED case, which does not use fall_back:. Tested-by: Steve Dickson <steved@redhat.com> Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* mount.nfs: Restore errno after v3 mounts on ECONNREFUSED errorsSteve Dickson2017-05-051-1/+4
| | | | | | | | | | | | Commit 65ac59cd introduced code that tries v3 mounts when the the v4 mount fails with ECONNREFUSED. This code allows failing back to v3 to work correctly. When the v3 mount fails the original errno value has been over rewritten. In these case the errno value needs to be restored to ECONNREFUSED. Signed-off-by: Steve Dickson <steved@redhat.com>
* config: Remove the conf_path globalJustin Mitchell2017-05-0515-58/+45
| | | | | | | | Working towards an nfs.conf library and API for system config tools, first step, replace the conf_path global with a parameter to conf_init Signed-off-by: Justin Mitchell <jumitche@rehat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsdcltrack: silence some expected errorsJeff Layton2017-04-261-2/+2
| | | | | | | | | | | On a new install, we're unable to select from the parameters table, as it doesn't exist yet. The code is set up to log that fact at L_ERROR now, but it's an expected situation. Change it to log that at D_GENERAL instead. Reported-and-Tested-by: ChunYu Wang <chunwang@redhat.com> Signed-off-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* systemd: Afters are also needed for the Wants=network-online.targetSteve Dickson2017-04-244-3/+4
| | | | | | | | | | | Commit 9d4fc3fb added Wants=network-online.target which is not enough to ensure the network is completely up before the NFS server is started. After=network-online.target is also needed. Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1419351 Signed-off-by: Steve Dickson <steved@redhat.com>
* systemd: NFS server services should use network-onlineSteve Dickson2017-04-104-5/+6
| | | | | | | | | | | | | | | | There has been an number startup problems where parts of the NFS server fails to start due to DNS and other parts of the network not be up. Reading the systemd.special it seems network.target is a passive unit which does not wait for the entire network to come up and network-online.target is an active unit which does wait. So this adds Wants=network-online.target to all of the NFS server services Signed-off-by: Steve Dickson <steved@redhat.com>
* systemd: add a generator for the rpc_pipefs mountpointScott Mayhew2017-04-1011-43/+311
| | | | | | | | | | | | | | | | | | | | | | | | The nfs.conf has config options for the rpc_pipefs mountpoint. Currently, changing these from the default also requires manually overriding the systemd unit files that are hard-coded to mount the filesystem on /var/lib/nfs/rpc_pipefs. This patch adds a generator that creates a mount unit file for the rpc_pipefs when a non-default value is specified in /etc/nfs.conf, as well as a target unit file to override the dependencies for the systemd units using the rpc_pipefs. The blkmapd, idmapd, and gssd service unit files have been modified to define their dependencies on the rpc_pipefs mountpoint indirectly via the rpc_pipefs target unit file. Since both rpc-pipefs-generator.c and nfs-server-generator.c need to convert path names to unit file names, that functionality has been moved to systemd.c. This patch also removes the dependency on the rpc_pipefs from the rpc-svcgssd.service unit file. rpc.svcgssd uses the sunrpc cache mechanism to exchange data with the kernel, not the rpc_pipefs. Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* blkmapd: allow the rpc_pipefs mountpoint to be overriddenScott Mayhew2017-04-093-11/+56
| | | | | | | | Allow the rpc_pipefs mountpoint to be overriden via the pipefs-directory variable in the [general] section of /etc/nfs.conf. Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* gssd: add a deprecation warning for pipefs-directory in gssd sectionScott Mayhew2017-04-094-6/+14
| | | | | | | | All the daemons should use the same rpc_pipefs, so pipefs-directory should be specified in the [general] section. Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* idmapd: move the pipefs-directory config option to nfs.confScott Mayhew2017-04-094-11/+60
| | | | | | | | | | | | | Changed idmapd to read its value for the pipefs-directory from /etc/nfs.conf rather than /etc/idmapd.conf. All other configurations related to id mapping still reside in /etc/idmapd.conf for now. Added a warning to indicate that idmapd's -c option is deprecated. Corrected a misspelling of 'configuration' in nfs.conf. Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* systemd: nfs-server service should use network-onlineSteve Dickson2017-04-091-2/+2
| | | | | | | | | | | | | | | There has been an number startup problem where parts of the NFS server fail to start due to DNS and other parts of the network not be up. Reading the systemd.special it seems network.target is a passive unit which does not wait and network-online.target is an active unit which does not wait so that should be used. Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1419351 Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: don't enable a UDP socket by defaultJeff Layton2017-04-054-13/+23
| | | | | | | | | | | | | | | | | | | Most major NFS clients have supported TCP for at least a decade now, and v4-only shops are becoming more prevalent. It seems reasonable that serving over UDP should be something that is "opt-in". I've always hesitated to do this in the past, but now that we have nfs.conf, it seems like the time may be right to disable UDP in default configurations. In particular, it would be good to try this in the more bleeding edge distros (Fedora, Ubuntu, SuSE, etc...) and see how problematic it is. Change the default in rpc.nfsd to just open TCP ports by default. Add new -u and -t options that allow users to explicitly override what's in the config file, and update the usage message and manpage. Signed-off-by: Jeff Layton <jlayton@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: Allow the caller to turn off NFSv4.0 without turning off NFSv4.xTrond Myklebust2017-04-055-35/+92
| | | | | | | | | | | | The new semantic is that '-N4' turns off all NFSv4 minor versions, while '-V4' turns them all on. In order to turn off just minor version x (x >= 0), use -N4.x, and to turn it back on. '-V4.x'. Note that on older kernels, attempting to use -N4.0 and -V4.0 is equivalent to specifying -N4 or -V4. Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* sm-notify: ending the grace period early should be configurableScott Mayhew2017-04-043-1/+24
| | | | | | | | | | | | | sm-notify's early ending of the grace period when it has no hosts to notify can cause problems in some high availability configurations, which may be running one sm-notify per floating IP address in the cluster. This commit makes that behavior configurable via the nfs.conf (I don't think having a corresponding command line option would be particularly useful, hence none was added). Acked-by: J. Bruce Fields <bfields@fieldses.org> Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* Manual page bug: two inaccuracies in nfsd(7)Chris Siebenmann2017-03-141-6/+2
| | | | | | | | | | | Here is a patch that deletes the mention of auth.domain and changes the wording around 'flush' files. I'm not attached to the revised wording; it's just the best I could manage in something that felt that it was within the same style and space as the current wording. Acked-by: NeilBrown <neilb@suse.com> Signed-off-by: Chris Siebenmann <cks.nfs01@cs.toronto.edu> Signed-off-by: Steve Dickson <steved@redhat.com>
* gssd: Duplicate the upcall string for error messagesFrank Sorenson2017-02-151-4/+13
| | | | | | | | | | | strsep() modifies the input string, so error messages may output only part of the upcall string. Make a copy of the upcall string, and use that in any error messages. Signed-off-by: Frank Sorenson <sorenson@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* gssd: replace non-thread-safe strtok with strsepFrank Sorenson2017-02-151-1/+2
| | | | | | | | | | gssd uses the non-thread-safe strtok() function, which can lead to incorrect program behavior. Replace strtok() with the thread-safe strsep(). Signed-off-by: Frank Sorenson <sorenson@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfs-server-generator: handle 'noauto' mounts correctlyNeilBrown2017-02-151-0/+26
| | | | | | | | | | | | | | | | | | | When this code was written, the systemd documentation stated that "RequiresMountsFor" ignored mountpoints marked as "noauto". Unfortunately this is incorrect. Consquently a filesystem marked as noauto that is also NFS exported will currently be mounted when the NFS server is started. This is not what people expect. So add a check for the noauto flag. If any ancestor of a given export point has the noauto flag, no RequiresMountsFor will be generated for that point. Also skip RequiresMountsFor for exports marked 'mountpoint', as their absence is, theoretically, already handled by mountd. URL: https://github.com/systemd/systemd/issues/5249 Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* mount: call setgroups() before setuid()NeilBrown2017-02-151-0/+2
| | | | | | | | | | | | | | | It is generally wise to call setgroups() (and setgid()) before calling setuid() to ensure no unexpected permission leaks happen. SUSE's build system checks all binaries for conformance with this and generates a warning for mountd. As we are setting the uid to 0, there is no risk that the group list will provide extra permissions, so there is no real risk here. But it is nice to silence warnings, and including a setgroups() call is probably a good practice to encourage. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* conffile: ignore empty environment variables.NeilBrown2017-02-151-1/+1
| | | | | | | | | | | | | | | | | conf_set() already refuses to set an empty value, so if foo= appears in the config file, it will be ignored. This patch extends the policy to environment variables, so empty environment variables are treated as though they didn't exist. This means that a separate environment file (e.g. /etc/sysconfig/nfs) will be treated the same way whether it is: - included in the [environment] section of /etc/nfs.conf - sourced by the shell before running code - sourced by the systemd EnvironmentFile directive. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* systemd: Fix nfs-mountd dependency on rpcbindRafael David Tinoco2017-02-151-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | Following commit 91da135f - it replaced "rpcbind.target" by "rpcbind.socket" in some unit files - "rpcbind.socket" should also be added to "nfs-mountd.service" as a dependency to avoid race conditions. Usually "rpcbind.socket" is either started as a "sockets.target" dependency, or as a dependency for "nfs-server.service", when unit files include it in "BindsTo" or "After". Unfortunately there is a possilibility to have "nfs-mountd.service" started when the rpcbind socket is not yet created: systemd[1]: Starting NFS Mount Daemon... systemd[1]: nfs-mountd.service: Control process exited, code=exited status=1 systemd[1]: Failed to start NFS Mount Daemon. systemd[1]: nfs-mountd.service: Unit entered failed state. systemd[1]: nfs-mountd.service: Failed with result 'exit-code'. Nowadays "nfs-mountd.service" uses "BindTo" directive to "nfs-server.service". That, per se, doesn't guarantee ordering for NFS server to start rpcbind and for nfs-mountd to depend on it. https://bugs.launchpad.net/bugs/1590799 Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Rafael David Tinoco <rafael.tinoco@canonical.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* mountd/exportfs: implement the -s/--state-directory-path optionScott Mayhew2017-02-1511-50/+188
| | | | | | Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* libnsm.a: refactor nsm_setup_pathnames() and nsm_make_pathname()Scott Mayhew2017-02-155-43/+119
| | | | | | | | | | Move the logic in nsm_setup_pathnames() and nsm_make_pathname() to similar generic functions in libmisc.a so that the exportfs and rpc.mountd programs can make use of them later. Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Scott Mayhew <smayhew@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* mount: fix mount fail that caused by uninitialized structJianhong Yin2017-02-023-11/+24
| | | | | | | | | | | | | From: "Jianhong.Yin" <yin-jianhong@163.com> recent changes of utils/mount cause a regression mount fail: https://bugzilla.redhat.com/show_bug.cgi?id=1415024 can not reproduce it on x86_64(gcc on x86_64 might do struct initialize by default, I'm not sure). but it can be reproduced always on platform ppc64le aarch64. Signed-off-by: Jianhong Yin <yin-jianhong@163.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* Release: 2.1.1Steve Dickson2017-01-121-1/+1
| | | | Signed-off-by: Steve Dickson <steved@redhat.com>
* exportfs: support "security_label" export optionJ. Bruce Fields2017-01-124-1/+16
| | | | | | | | On recent kernels only exports with NFSEXP_SECURITY_LABEL set will export security labels. Signed-off-by: J. Bruce Fields <bfields@redhat.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsstats: fix some exit codes.NeilBrown2017-01-111-3/+3
| | | | | | | | | | | "nfsstat -m" always exits with "1" (unless there is an error opening /proc/mounts). It should exist "0". Also, a few usage errors cause an exit of "255" when it should probably be "1". Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* Renamed nfs.config to nfs.confSteve Dickson2017-01-041-0/+0
| | | | Signed-off-by: Steve Dickson <steved@redhat.com>
* exportfs: Make sure pass all valid export flags to nfsdKinglong Mee2017-01-041-5/+7
| | | | | | | | | test_export pass a export flags only marks NFSEXP_FSID, nfsd may want other flags for export checking. This patch make sure exportfs pass all other flags to nfsd. Signed-off-by: Kinglong Mee <kinglongmee@gmail.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfs-server-generator: avoid using syslogNeilBrown2017-01-041-0/+3
| | | | | | | | | | | | nfs-server-generator is run very early when a lot of services are not yet started, so it mustn't depend on them. It already avoids using DNS, but it should avoid syslog too. If it tries to log error to syslog, it can deadlock. So just let messages go to stderr. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* mountd: delay reading etab until first request arrives.NeilBrown2017-01-041-2/+0
| | | | | | | | | | | | | | | Reading etab may require hostname lookup, so it is not reliable until the network is active. But we want mountd to start before that so that it is ready when the very first NFS request arrives. So delay reading etab until that request arrives, by which time the network must be online so hopefully hostname look will be reliable. An alternate would be to delay starting mountd and nfsd until the network is on-line, but that will often be an unnecessary delay. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: Do not permit manipulation of NFSv4.0, e.g. "-N 4.0"NeilBrown2017-01-042-4/+4
| | | | | | | | | The code maps this into "-4.32", which the kernel rejects. The kernel also rejects "-4.0" (when written to the 'versions' file). So require the minor number to be at least NFS4_MINMINOR, which is '1'. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: fix setting of minor version from config fileNeilBrown2017-01-042-5/+17
| | | | | | | | | | | | | | Several problem here: - code didn't actually work, as it cleared a bit from minorversset when it should have cleared from minorvers - code did not allow minor versions to be enabled, which is useful when a new minor version is partially implemented in the kernel but not yet enabled by default - code allowed version 4.0 to be enabled/disabled, which the kernel does not support (as for 4.9 at least). Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfs.conf: adding a reference file of /etc/nfs.configSteve Dickson2016-12-201-0/+70
| | | | Signed-off-by: Steve Dickson <steved@redhat.com>
* svcgssd: add /etc/nfs.conf supportNeilBrown2016-12-203-0/+40
| | | | | Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* gssd: add /etc/nfs.conf supportNeilBrown2016-12-203-0/+101
| | | | | Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* exportfs: allow debugging to be enabled in nfs.confNeilBrown2016-12-203-0/+19
| | | | | Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* xlog: Add common support for "debug=??" in /etc/nfs.confNeilBrown2016-12-208-1/+29
| | | | | | | | The value is from the list general, call, auth, parse, all. Most daemons recognise this in their dedicated section. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* conffile: fix striping of quotes from values.NeilBrown2016-12-201-8/+8
| | | | | | | There were being stripped from the name instead! Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* conffile: allow embedded spaces in values.NeilBrown2016-12-201-2/+6
| | | | | | | | | | | | | | | | The code that said "Skip trailing spaces" actually skipped everything after the first space. Change to to only skip trailing spaces, or comments that start after a space. This is useful for lists: Foo: a, b, c The list handling already allows for internal spaces. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: add /etc/nfs.conf support for host=NeilBrown2016-12-203-0/+23
| | | | | | | | | A comma separate list of hosts can be given. If any host name is given with "--host" or "-H", then all hosts listed in nfs.conf are ignored. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: add /etc/nfs.conf support for nfsd.port option.NeilBrown2016-12-203-2/+10
| | | | | Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: remove pointless memory allocations.NeilBrown2016-12-201-9/+4
| | | | | | | | There is not need to e.g. strdup(optarg), and the value is constant. It can just be used directly. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* nfsd: move and improve test on valid portNeilBrown2016-12-201-7/+10
| | | | | | | | | | | | | | | nfssvc_set_sockets() access textual port numbers (by lookup in /etc/services). This uses getaddrinfo which reports errors, except for out-of-range numbers. So change the test on a valid port to only complain if the port given is purely numeric, but is out-of-range. Also move it so that any default value gets tested the same as any argument value. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* Add nfs.systemd man pageNeilBrown2016-12-202-2/+169
| | | | | | | | | This discusses some of the behaviors of the various unit files, and how best to work with them to achieve various results. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* systemd: Remove the nfs-config.serviceNeilBrown2016-12-2012-66/+23
| | | | | | | | | | | | | | | | | | | | | | | | | | | Now that we have /etc/nfs.conf, a lot of configuration can be read directly. So nfs-config isn't really needed any more. Some distributions allow command-line arguments for various daemons to be set in an environment file (/etc/sysconfig, /etc/defaults). Passing these through /etc/nfs.conf is not possible. Instead, a distro that needs this functionality can create drop-in files which select the required value. As no commands are given default arguments by systemd unit files, the drop-in can just add distro-specific args. For example /lib/systemd/system/nfs-mountd.service.d/local.conf [Service] EnvironmentFile=/etc/sysconfig/nfs ExecStart= ExecStart=/usr/sbin/rpc.mountd $RPCMOUNTDOPTS Note the need for the empty assignment to remove existing definitions first. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* statd: allow --no-notify to be passed via environment variable.NeilBrown2016-12-203-1/+12
| | | | | | | | | | The significant value of allowing this is that it means that for default operation, systemd unit files do not need to pass any options to any programs. The purpose of this will become apparent in the next patch. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
* conffile: allow $name expansion of tag values.NeilBrown2016-12-202-2/+22
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | If the value for a tag starts with '$', then the remainder of the value is treated as an environment variable name. It is looked up in the environment (getenv) and if not found, it is looked for in the [environment] section of the config file. This lookup is formed as access time e.g. by conf_get_str(), not at parse time. The expected usage is that the config file can contain something like [environment] include = /etc/sysconfig/nfs [other-section] tag = $NAME and conf_get_str("other-section","tag") will report the value of "NAME" in the given file. As different distributions used different environment files, and different variable names with-in them, a distro could provide a static config file which maps from names in that environment file to config tags requires by NFS daemons. Signed-off-by: NeilBrown <neilb@suse.com> Signed-off-by: Steve Dickson <steved@redhat.com>
generated by cgit (git 2.34.1) at 2025-09-15 09:51:41 +0000