On 10/30/2013 05:30 PM, Eric Blake wrote:
Actually put gfapi to use, by allowing the creation of a gluster pool. Right now, all volumes are treated as raw; further patches will allow peering into files to allow for qcow2 files and backing chains, and reporting proper volume allocation.
Needs a v2 for several reasons:
+ +static void +virStorageBackendGlusterClose(virStorageBackendGlusterStatePtr state) +{ + if (!state || !state->vol) + return; + /* Yuck - glusterfs-api-3.4.1 appears to always return -1 for + * glfs_fini, with errno containing random data, so there's no way + * to tell if it succeeded. 3.4.2 is supposed to fix this.*/ + if (glfs_fini(state->vol) < 0) + VIR_DEBUG("shutdown of gluster failed with errno %d", errno); +}
Leaks state.
+ +static virStorageBackendGlusterStatePtr +virStorageBackendGlusterOpen(virStoragePoolObjPtr pool) +{ + virStorageBackendGlusterStatePtr ret = NULL; + + if (VIR_ALLOC(ret) < 0) + return NULL; + + if (!(ret->vol = glfs_new(pool->def->source.name))) {
Fails for subdirectory access (qemu allows gluster://host/vol/dir/file, so we should allow vol/dir as a pool name in the XML, but glfs requires us to open vol then chdir to dir, rather than opening vol/dir in one go).
+ virReportOOMError(); + goto error; + } + + /* FIXME: allow alternate transport in the pool xml */ + if (glfs_set_volfile_server(ret->vol, "tcp", + pool->def->source.hosts[0].name, + pool->def->source.hosts[0].port) < 0 || + glfs_init(ret->vol) < 0) { + virReportSystemError(errno, _("failed to connect to gluster %s/%s"), + pool->def->source.hosts[0].name, + pool->def->name);
Reports the wrong name (should be pool->def->source.name). And I'm still working on the followup patches for qcow2 metadata parsing... -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org