| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
If authentication is required to use virsh, sos will timeout on
'virsh list --all'. Use '-r' option to avoid this.
2016-02-10 00:24:02,761 WARNING: [plugin:virsh] command 'virsh list
--all' timed out after 300s
$ virsh list --all
Please enter your authentication name:
Please enter your password:
error: failed to connect to the hypervisor
error: no valid connection
error: authentication failed: Failed to step SASL negotiation: -1
(SASL(-1): generic failure: All-whitespace username.)
$ virsh -r list --all
Id Name State
----------------------------------------------------
Fixes: #761.
Signed-off-by: Germano Veit Michel <germano@redhat.com>
|
|
|
|
|
|
|
| |
Closes:#730
Signed-off-by: Amit Ghadge <amitg.b14@gmail.com>
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
| |
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
If a profile has no usable plugin do not add its profiles to the
list of available profiles and do not include them in the output
of --list-profiles or --list-plugins.
Fixes: #764.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
| |
There's no point having both callers bracket the call with lines
that output a single blank line: do this all in the method itself.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When sosreport prints a profile listing as a result of the user
selecting a non-existent profile we currently output a profile
list (no plugin names) that ends with the standard summary of
profiles and plugins:
# sosreport -vv --batch --profile qux
set sysroot to '/' (default)
sosreport (version 3.2)
Unknown or inactive profile(s) provided: qux
The following profiles are available:
boot
cluster
[...]
system
virt
webserver
28 profiles, 0 plugins
This is less than useful and confusing for users (who shouldn't
have to care about how we manage this stuff internally).
Change this to only include the ", N plugins" part of the text
if we have actually loaded some valid plugins for the active
profile and factor out common display code in list_plugins and
list_profiles.
Fixes: #733.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
The 'netstat -agn' command reports IPv4 and IPv6 group addresses
which may also be truncated - apply the -W/-T switch to this
command as well.
Related: #734.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
By default netstat will truncate addresses that are wider than
the current column width. Disable this to ensure that all IPv6
addresses can be captured in full.
On Red Hat a non-upstream option (-T) was added to versions
prior to 2.0 - handle these version checks in the RedHatNetworking
class and set the appropriate option for each.
Fixes #734.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Correct profiles = ('system',)
Closes:#747
|
|
|
|
|
|
|
|
| |
Fix the summary field to not contain embedded newlines and work
around a bug in distutil's handling of gzipped man pages in
setup.py.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Fix the version string in the sos package __init__.py.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add a basic policy class for SuSE distributions and implement a
simple OpenSuSE policy (currently this will work with any SuSE
release).
A new SuSEPlugin tagging class is introduced for SuSE-specific
plugin cases. The policy currently also allows the RedHatPlugin
tagging class to run: this quickly enables a large number of
plugins for the distribution but will require manual testing
to identify cases where a dedicated SuSEPlugin must be created.
Fixes #523.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Resolves #731, #732.
Signed-off-by: Pavel Moravec <pmoravec@redhat.com>
|
|
|
|
|
|
| |
Fixes: #726.
Signed-off-by: Abhijeet Kasurde <akasurde@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
The nmcli command may print warning messages if the version of the
tool does not match the running daemon: filter these warnings out
of the list of connections or devices to check.
Fixes: #710.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Resolves: #712, #714
Signed-off-by: Pavel Moravec <pmoravec@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
Cisco CDP products include a kernel module that presents several
large pseudo files in /proc and /sys. Blacklist these locations
to avoid attempting to read the content into reports.
Fixes #740.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When the global verify switch was introduced the behaviour of the
historical 'rpm.rpmva' plugin option was made to depend on the
global setting: without --verify no verification is carried out
even though it is clear this is not what the user intended.
Allow verification to take place in the rpm module if the rpm.va
option is True, regardless of the setting of the global --verify
option.
Fixes #735.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Closes: #739.
Signed-off-by: Abhijeet Kasurde <akasurde@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
More cert information is captured, the "ipa-replica-manage"
commands removed because they would timeout because
a password was needed, sanitize some collected data,
and some logs with wildcards were not properly collected
is now fixed.
Fixes: #754, #755.
Signed-off-by: Shane Bradley <sbradley@redhat.com>
|
|
|
|
|
|
|
|
|
| |
The output from 'omreport chassis biossetup' identifies a system's power
profile, which is very handy when troubleshooting performance issues.
Closes: #725
Signed-off-by: Bill Yodlowsky <byodlows@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
Add plugin to collect logs and hardware health from Dell OpenManage
Server Administrator (OMSA).
Dell OMSA is an OS application used to manage Dell PowerEdge Server
hardware.
Signed-off-by: Charles Rose <charles_rose@dell.com>
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Add a plugin to collect hardware data from IPMI-enabled
hosts:
ipmitool sel info
ipmitool sel log
ipmitool sensor list
ipmitool chassis status
ipmitool fru print
ipmitool bmc info
ipmitool sdr info
Fixes: #756.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Resolves: #750
Signed-off-by: Pavel Moravec <pmoravec@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The installed-rpms file is intended as a quick list of all installed
packages for distributions using the RPM package manager: the contents
have been fairly stable over several major releases and are formatted to
be human-readable in a standard 80x25 terminal without needing to filter
or sort.
The addition of several large fields over the last year has led to some
significant bloat however:
python-ply-3.6-2.fc23.noarch Thu Oct 8 10:55:19 2015 1444298119 Fedora Project buildvm-09.phx2.fedoraproject.org 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 RSA/SHA256, Tue Aug 18 14:51:44 2015, Key ID 32474cf834ec9cba
vs.
python-ply-3.6-2.fc23.noarch Thu Oct 8 10:55:19 2015
Revert the installed-rpms file back to the original format and add a new
file, package-data in the rpm plugin's commands directory that contains
these additional rpm query fields.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Historically, when calling an external process, sos relied on the
use of a shell to perform glob expansion on arguments passed to
the command.
This potentially exposes sos and its children to the entire gamut
of shell expansion and all the problems associated with it and
this is part of the reason that use of a shell in callouts was
disabled in commit 46b6c3d (Issue #253).
While full shell syntax is undesirable for many reasons the use
of globs to express path arguments to external commands is very
useful and widely applied: restore this limited support for
shell-like syntax using the python glob.glob module and expand
any argument that is a valid glob into its resulting arguments.
Fixes: #752
Signed-off-by: Shane Bradley <sbradley@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
The archive classes are not special snowflakes: they do not need
their own (buggy) version of sos_get_command_output(): replace the
hand-rolled Popen() call with the sos.utilities version.
Really Fixes: #742.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|\
| |
| | |
[redhat] A managled rhn systemid will crash sosreport.
|
|/
|
|
|
|
| |
Fixes: #742
Signed-off-by: Shane Bradley <sbradley@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
| |
This changes database dump collection format to 'list' which is a more
readable and easy to parse.
Based on a change suggested by Prasad Mukhedkar.
Resolves: #676.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
| |
Signed-off-by: Abhijeet Kasurde <akasurde@redhat.com>
|
|
|
|
|
|
|
| |
Ensure the correct path (in the system temporary directory) is
reported when the --build option is used.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
OpenShift broker and console can be configured with different
authentication backends. If LDAP is being used with a server that
requires authentication for search the configuration files will
capture the credentials of the BindDN.
Fixes: rhbz#1227462
Fixes: #673.
Signed-off-by: Pep Turró Mauri <pep@redhat.com>
|
|
|
|
|
|
| |
Fixes: #671.
Signed-off-by: Martin Frodl <mfrodl@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Ovirt AAA_JDBC stores database password in plain text under
/etc/ovirt-engine/aaa/*.properties file.
It contains password in the format:
config.datasource.dbpassword=<password>
The engine PostgreSQL user password and the custom build domain's user
password exist in plain text format in these files. The commit removes
these password from the sosreport.
Fixes: #666.
Signed-off-by: Nijin Ashok <nashok@redhat.com>
|
|
|
|
| |
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
| |
Collect output of "apachectl -M" for that
Resolves: #691
Signed-off-by: Pavel Moravec <pmoravec@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For more information about these parameters, see the md man page:
http://linux.die.net/man/4/md
The kernel.org documentation also has some information about these parameters:
https://www.kernel.org/doc/Documentation/md.txt
Fixes #700.
Signed-off-by: Kenneth Koski <mechaxl@gmail.com>
|
|
|
|
| |
Signed-off-by: Hisanobu Okuda <hokuda@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
Some openstack installations split out the configuration files
via an "include conf.d/*.cfg" directive. Add the whole directory
in order to collect that as well.
Fixes: #674.
Signed-off-by: Michele Baldessari <michele@acksyn.org>
|
|
|
|
| |
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
| |
Resolves: #697.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
To avoid file creation races in shared temporary directories like
/tmp and /var/tmp use a private (0700) subdirectory to build the
FileCacheArchive and subsequent archive and compressed archive
files: only create a file in the containing directory when it can
be done as a single atomic rename.
This prevents sos from writing to an arbitrary location under the
control of another user: a malicious user could steal data or over
write files in /etc resulting in a local privilege escalation.
There remains a further race since once the archive name is known
the checksum file name becomes predictable: as the checksum file
is also prepared in the subdirectory and moved into place the
result is always either success or an error that is reported to
the user.
The correct checksum value is still reported to the user via the
terminal.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Although the digest algorithm is policy controlled the actual
mechanism to checksum the archive does not belong in the policies
module: historically this was done to keep the code that calculates
the checksum close to the UI code that reports it.
Move the calculation to the main SoSReport class's final_work()
method and add a 'checksum' argument to the display_results()
method so that the value can be reported.
In future it may make sense to push the checksum code directly into
the archive class.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Pass explicit archive and build directory arguments to the
Policy.display_results() method rather than a single path name
argument and a boolean to indicate whether it is an archive file
or a build directory path.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
It's crazy having the Policy classes call a function in the
utilities module only to have that function then load the Policy
module, call policy.get_preferred_hash_algorithm() and then test
the result.
Get rid of the get_hash_name() function in the utilities module
and simplify the calls in the policies module to obtain the
preferred hash name.
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
| |
Signed-off-by: Bryn M. Reeves <bmr@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
Collect the two directories without some forbidden paths.
Reorder self.add_forbidden_path(..) to have some logical ordering there.
Resolves: #693
Signed-off-by: Pavel Moravec <pmoravec@redhat.com>
|
| |
|
| |
|