diff options
| -rw-r--r-- | README | 5 | ||||
| -rw-r--r-- | src/README | 15 | ||||
| -rw-r--r-- | src/TODO | 2 |
3 files changed, 20 insertions, 2 deletions
@@ -51,3 +51,8 @@ tools: development/debugging: fakesyn -- simulator with synthetic workload fakefuse -- simulator that mounts with fuse + +---- + +see also src/README + diff --git a/src/README b/src/README new file mode 100644 index 00000000000..3f897f7d18f --- /dev/null +++ b/src/README @@ -0,0 +1,15 @@ +A few helper scripts: + +mkcephfs.sh -- initialize a new, empty ceph filesystem (by creating a + fresh monfs) + +mkfs.sh -- same, but also initialize empty object stores. assumes + all osds are on the same node, with dev/osd# for storage. + +stop.sh -- stop/kill all daemons. + +restart.sh -- stop everything, then restart everything. (does not mkfs.) + +startnew.sh -- stop everything, initialize new ceph fs, and start it. + assumes everything is on one node. for ease of testing. + @@ -29,8 +29,6 @@ kernel client - do we need to block? - timeout mds session close on umount - deal with CAP_RDCACHE properly: invalidate cache pages? -- what happens after reconnect when we get successful reply but no trace (!) on e.g. rename, unlink, link, open+O_CREAT, etc... - - fallback in each case (in ceph_unlink, _rename, etc.) when reply has no trace? - procfs/debugfs - adjust granular debug levels too - should we be using debugfs? |