Comment #63 on issue 520135 by
diego.ca...@gmail.com: Update failed (error:
Boris, I was able to run the command which returned:
<?xml version="1.0" encoding="UTF-8"?><gupdate
xmlns="
http://www.google.com/update2/response" protocol="2.0"
server="prod"><daystart elapsed_days="3282" elapsed_seconds="63115"/><app
appid="com.google.Keystone" cohort="" cohortname="" status="ok"><ping
status="ok"/><updatecheck DisplayVersion="1.2.3.738"
codebase="
http://dl.google.com/dl/keystone/googledummysoftwareupdate-1.2.3.738.dmg"
fp="1.903b5331c4ebe1cb8b32b616e9ffce13c3d3d46d114b8e3aba24aa3167d675c0"
hash="M0pkhPp0MQsGhybA+uQuVMC0WwI="
hash_sha256="903b5331c4ebe1cb8b32b616e9ffce13c3d3d46d114b8e3aba24aa3167d675c0"
needsadmin="false" size="734780" status="ok"/></app></gupdate>
Notwithstanding this result, Chrome is still stuck with the update.
May not be of help, but tried to kill ksadmin while the Chrome update tab
was open: it got respawned but nothing else, instead, after closing the
tab, ksadmin spawned ksfetch.
Then if I keep the tab closed, both processes go away after a number of
minutes (didn't measure) - looking at the system log, ksfetch timed out and
failed.
The sequence is repeatable.
While looking at the system log, I've noticed a few weird things like:
- ksfetch timeout errors are always related only to Chrome
- apparently, Drive updates work (I don't have enough knowledge about ks,
but the update XML is there and definitely similar to what you posted)
- KSAgentApp can successfully upload ks stats
- I've found the Keystone.ticketstore.lock file in
/Library/Google/GoogleSoftwareUpdate/TicketStore/ and since it was quite
old I thought it was stale and tried removing it, but without any luck:
-rw-r--r-- 1 root wheel 1827 16 Dic 13:34 Keystone.ticketstore
-rw-r--r--@ 1 root wheel 0 13 Nov 17:43 Keystone.ticketstore.lock
- there's also an error (by Drive) complaining about not being able to get
a lock on /tmp/.keystone_install_lock, which again seems stale
-rw-rw-rw- 1 diego wheel 0 9 Dic 10:34 .keystone_install_lock
- in /tmp I've found also folders like:
drwx------ 3 root wheel 102 28 Dic 01:18
KSOutOfProcessFetcher.0.sAglCyxY5lzPoNgfmEvv-ZqGl-w=
drwx------ 3 diego wheel 102 28 Dic 02:01
KSOutOfProcessFetcher.501.sAglCyxY5lzPoNgfmEvv-ZqGl-w=
both contain only the ksfetch executable, both same length but the one
belonging to my user has the little @ sign at the end of the ls -l command:
-rwx------@ 1 diego wheel 163472 28 Dic 02:01 ksfetch
which in the OSX world means extended attributes, so looking at them prints:
$ sudo xattr -l
KSOutOfProcessFetcher.501.sAglCyxY5lzPoNgfmEvv-ZqGl-w\=/ksfetch
com.apple.quarantine: 0000;568097fd;Google Chrome.app;
The com.apple.quarantine attribute is related to the automatic blocking of
execution that OSX applies to files downloaded from the Internet; after
removing the attribute, apparently anything didn't change, looking at /tmp
instead I've found 4 folders related to ks, the new ones containing
a "download" executable:
drwx------ 3 root wheel 102 28 Dic 02:23
KSOutOfProcessFetcher.0.sAglCyxY5lzPoNgfmEvv-ZqGl-w=
drwx------ 3 diego wheel 102 28 Dic 02:23
KSOutOfProcessFetcher.501.sAglCyxY5lzPoNgfmEvv-ZqGl-w=
drwx------ 3 root wheel 102 28 Dic 02:23
KSOutOfProcessFetcher.bFHhEmnEsS
drwx------@ 3 diego wheel 102 28 Dic 02:23
KSOutOfProcessFetcher.q2Aey4k8hJ
As you can see, the last one has the @ sign which again means
com.apple.quarantine, which is also attached to the download file in it,
and to the ksfetch on the folder owned by my user (which makes sense
keeping into account root and normal user behavior of OSX).
Chrome update still didn't work and, after removing all the folders, error
11 is still there.
I've tried disabling Gatekeeper from System Preferences (Allow apps
downloaded from anywhere) with no luck (after closing and reopening Chrome).
Apparently all this leads to the obvious conclusion that Chrome/ks have a
bad relation with OSX/Gatekeeper, which seems weird to me but who knows,
this makes definitely sense as it explains why ksfetch times out and
matches your suspect about an antivirus issue.