)]}'
{"ocf/NovaCompute":[{"author":{"_account_id":2394,"name":"Adam Spiers","email":"aspiers@suse.com","username":"adam.spiers"},"change_message_id":"1b263b66c7beb1b1615b6599af1101d23194ed2c","unresolved":false,"context_lines":[{"line_number":161,"context_line":""},{"line_number":162,"context_line":"    state\u003d$(attrd_updater -p -n evacuate -N ${NOVA_HOST} | sed -e \u0027s/.*value\u003d//\u0027 | tr -d \u0027\"\u0027 )"},{"line_number":163,"context_line":"    if [ \"x$state\" \u003d x ]; then"},{"line_number":164,"context_line":"\t: never been fenced"},{"line_number":165,"context_line":""},{"line_number":166,"context_line":"    elif [ \"x$state\" \u003d xno ]; then"},{"line_number":167,"context_line":"\t: has been evacuated, however it could have been 1s ago"}],"source_content_type":"application/x-shellscript","patch_set":1,"id":"dae33548_bc963c92","line":164,"range":{"start_line":164,"start_character":0,"end_line":164,"end_character":1},"updated":"2016-02-15 14:36:28.000000000","message":"I just realised after merging this that it changes the previous assumption that the absence of an attribute means the node has never been fenced.  So this \"never been fenced\" line is now wrong, and I wonder whether it\u0027s possible for the state to be deleted within an unlucky window resulting in the sleep ${OCF_RESKEY_evacuation_delay} being skipped.  What do you think?","commit_id":"2fffcca22828c02dfe280c40fdad933a8a0fc270"},{"author":{"_account_id":2394,"name":"Adam Spiers","email":"aspiers@suse.com","username":"adam.spiers"},"change_message_id":"2ae67a78bf926090c70c34d504cae022360af4f9","unresolved":false,"context_lines":[{"line_number":161,"context_line":""},{"line_number":162,"context_line":"    state\u003d$(attrd_updater -p -n evacuate -N ${NOVA_HOST} | sed -e \u0027s/.*value\u003d//\u0027 | tr -d \u0027\"\u0027 )"},{"line_number":163,"context_line":"    if [ \"x$state\" \u003d x ]; then"},{"line_number":164,"context_line":"\t: never been fenced"},{"line_number":165,"context_line":""},{"line_number":166,"context_line":"    elif [ \"x$state\" \u003d xno ]; then"},{"line_number":167,"context_line":"\t: has been evacuated, however it could have been 1s ago"}],"source_content_type":"application/x-shellscript","patch_set":1,"id":"dae33548_5f49e28a","line":164,"range":{"start_line":164,"start_character":0,"end_line":164,"end_character":1},"in_reply_to":"dae33548_5c5f3072","updated":"2016-02-15 15:07:56.000000000","message":"Yes please, would be good to tweak that.\n\nYou\u0027re right that this doesn\u0027t introduce any new race window.  But your other patch Id981dc1b80bbbac7b294ab7fd9e24b3757765826 which deletes attributes might.  Why did you introduce the deletion?  I didn\u0027t think carefully enough about that before merging it.","commit_id":"2fffcca22828c02dfe280c40fdad933a8a0fc270"},{"author":{"_account_id":4460,"name":"Vincent Untz","email":"vuntz@opensuse.org","username":"vuntz"},"change_message_id":"09df6c8218acdd280bad86a0304ab7beece6555f","unresolved":false,"context_lines":[{"line_number":161,"context_line":""},{"line_number":162,"context_line":"    state\u003d$(attrd_updater -p -n evacuate -N ${NOVA_HOST} | sed -e \u0027s/.*value\u003d//\u0027 | tr -d \u0027\"\u0027 )"},{"line_number":163,"context_line":"    if [ \"x$state\" \u003d x ]; then"},{"line_number":164,"context_line":"\t: never been fenced"},{"line_number":165,"context_line":""},{"line_number":166,"context_line":"    elif [ \"x$state\" \u003d xno ]; then"},{"line_number":167,"context_line":"\t: has been evacuated, however it could have been 1s ago"}],"source_content_type":"application/x-shellscript","patch_set":1,"id":"dae33548_5c5f3072","line":164,"range":{"start_line":164,"start_character":0,"end_line":164,"end_character":1},"in_reply_to":"dae33548_bc963c92","updated":"2016-02-15 14:47:54.000000000","message":"You\u0027re right, it should now be a \"no fencing to recover\" comment. Want me to submit a patch for this?\n\nI\u0027m not sure about the unlucky window you mention, though -- if that attribute is deleted by the user, then I can imagine you\u0027re correct, but that\u0027s really no different from before :-)\n\nIf it\u0027s deleted by this script, then the script always does it after the sleep (not done in this patch, but done in another patch you just merged, iirc).","commit_id":"2fffcca22828c02dfe280c40fdad933a8a0fc270"}]}
