Commit Graph

324 Commits

Author SHA1 Message Date
Digimer
1498e1b53c * Got server migration working using ocf:alteeve:server in a test environment!
* Converted most 'eval { }' calls to localize $@ and test the output of the eval, instead of checking to see if $@ was set.
* Converted all 'local' hash references to instead use the short host name of the local machine as a new standard.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-19 18:54:09 -04:00
Madison Kelly
30f2b3fa8e * Switched all hash 'local' keys to be the host's short user name. Untested, likely bugs to be fixed in the next commit.
Signed-off-by: Madison Kelly <mkelly@alteeve.ca>
2020-08-18 19:34:08 -04:00
Digimer
47203490a9 * Working on getting live migration to work with ocf:anvil:striker using the environment variables that pacemaker sets. Incomplete, but getting close.
* Added support to Cluster->parce_cib() to track if maintenance mode is set.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-17 11:55:49 -04:00
Digimer
e35800c413 * Fixed up (though more testing/work needed) to ocf:alteeve:server to get it working with DRBD resources referenced using '/dev/drbd/by-res/...'.
* Added the anvil.conf option 'sys::privacy::strong' that controls if the Anvil! ever "calls home". Initially, this controls DRBD's usage flag.
* Updated DRBD->get_devices() to track resources by their 'by-res' names as well and by the normal '/dev/drbdX' devices.
* To mitigate https://bugzilla.redhat.com/show_bug.cgi?id=1868467, updated Get->bridges() to parse the normal (non-JSON) data if we get invalid JSON output.
* Updated anvil-join-anvil to not disable, and in fact enable, libvirtd on boot. With DRBD 9, the original fear of a user accidentally booting a VM that's running on the peer no longer is an issue. By enabling it and leaving it on, Striker dashboard users won't lose their virtual machine manager link unless the node powers off. Also enabled actually updating the job progress, completing this tool!

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-13 00:12:20 -04:00
Digimer
39b4a912af * Remember in the last commit how I said that DRBD->update_global_common() was done? Well that was cute, 'cause it was quite broken. Now it's working.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-10 15:42:05 -04:00
Digimer
d647014ad1 * Created (finished but not yet tested) DRBD->update_global_common() to update DRBD's global_common.conf file.
* Cleaned up a lot of log levels.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-07 23:12:11 -04:00
Digimer
ef208fd3fb * Finished the logic for adding stonith devices and levels to pacemaker! More testing is needed though, bugs expected, but it adds them.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-06 17:45:43 -04:00
Digimer
c27cc7507f * Renamed striker-parse-fence-agents to anvil-parse-fence-agents and changed anvil-daemon to run it on all machines.
* Cleaned up a lot of logging.
* Updated Cluster->parse_cib() to track if a stonith device has 'delay' set.
* Got a lot more work done on anvil-join-anvil's stonith processing, but it still isn't complete. Updated it to change shell user passwords as well.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-08-06 01:20:53 -04:00
Digimer
61f4dcc41f * Updated Cluster->parse_cib() to pull out fencing (stonith) devices and levels.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-31 01:45:20 -04:00
Digimer
3c2f25a860 * Added 'fence_delay' fence agent to handle the corner cases where an IPMI BMC had crashed until a power cycle, and PDU fencing was effected, but failed to report as such.
* Updated Cluster->parse_cib() to take a CIB as a parameter.
* Fixed a bug in Database->get_hosts() where loading the host_ipmi value was filtered through Log->is_secure.
* Updated Striker->get_fence_data() to parse the switches to make it easier to map a fence agent's command line switches to STDIN arguments.
* Created System->parse_arguments() that converts a series of command line switches and their values into a hash. It's similar to Get->switches(), but works on any string.
* Continued work on anvil-join-anvil's fence configuration logic.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-30 00:23:47 -04:00
Digimer
d2d5d7b460 * Fixed a bug in Striker->load_manifest() where fences were parsed twice, the second time missing a hash reference.
* Updated striker to now only offer gateway for IFN networks. EL8 seems to ignore 'GATEWAY="x"' in interface configs which caused anvil-join-anvil to always think an interface needs to be updated. Updated as well to remove DNS entries set in interfaces that are not the default gateway.
* Fixed a bug where DNS entries were being missed, causing entries to be repeatedly added to the interface that was the gateway interface.
* In anvil-update-states, added Get->switches() so that verbosity switches are used.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-28 00:59:02 -04:00
Digimer
1bf71f8428 * Updated Database->get_hosts() to run host_ipmi the Log->is_secure if the string contains 'passw'.
* Fixed Database->get_ip_addresses() to clear stale IP addresses.
* Finished (for now, more testing needed) System->configure_ipmi! Also created System->test_ipmi() that handles trying lanplus and various password lengths, updating hosts -> host_ipmi on successful check.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-24 16:22:27 -04:00
Digimer
ef70f90ba4 * Updated Log->entry() to set the log file handle to UTF-8 when opened.
* Got more work done to System->configure_ipmi() to warm reset HP IPMI BMCs. It also now finds the IPMI user have started the password management.
* Created Words->shorten_string() that shortens a string to a number of bytes (as opposed to shortening to a character length).

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-22 22:32:33 -04:00
Digimer
dcfdf1127c * Got more work done on System->configure_ipmi(). It should now configure the IP address, subnet mask and default gateway using information from the manifest and anvil-join-anvil data.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-22 03:00:44 -04:00
Digimer
1fa63d2ea3 * Added 'anvil_uuid' as a set parameter in Database->get_hosts().
* Added calling 'debug => $debug' in System->X methods.
* Got more work done on System->configure_ipmi(). It should now determine if a BMC exists and pull the OEM and network details automatically.
* Updated anvil-configure-host to log more data in an attempt to find a reproducer for an odd bug where (apparently) a host was picking up the wrong job data meant for another host.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-21 02:00:56 -04:00
Digimer
99afd2e936 * Fixed a bug in Database->manage_anvil_conf() where initializing a host set the DB information with the wrong DB port and password.
* Fixed a bug in Get->host_type() the type wasn't being set for nodes and dr hosts.
* Fixed a bug in Validate->host_name() where the wrong method was being called.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-17 22:49:25 -04:00
Digimer
345d2e33d4 * Updated Cluster->parse_cib() to pre-fill some hashes to avoid undefined errors.
* Updated Storage->rsync() to only create the rsync wrapper if a password was given, allowing for rsync to work to/from a remote system when passwordless SSH is enabled.
* Updated anvil-join-anvil to disable/stop drbd.service, and to properly wait until both nodes are online.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-16 12:40:53 -04:00
Digimer
01974d7efe * Finished (though testing is needed) the updated ocf:alteeve:server resource agent. It now handles starting and stopping libvirtd and drbd daemons on-demand.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-16 00:33:00 -04:00
Digimer
dcd1fd1492 * Created Cluster->check_node_status() that checks the status of a node (in pacemaker).
* Created Cluster->get_peers() that figures out who the peer node (and DR host, if applicable) are.
* Updated Cluster->parse_cib() to dig out more information.
* Created Cluster->start_cluster() to start pacemaker (via pcsd) locally or on all (both) nodes.
* Started working on ocf:alteeve:server to start/stop the libvirtd/drbd daemons as needed, instead of having pacemaker do it.
* Got more work done on anvil-join-anvil. Node 2 now waits for the cluster to start, and node 1 will do setup as needed, then wait for the cluster to start as well.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-15 18:40:37 -04:00
Digimer
62d0a2aa39 * Created Cluster->parse_cib() that parses pacemaker's CIB (cluster information base) XML. This also switches to the XML::LibXML, starting the replacement of XML::Simple. It's far from finished, but parses out basic node data and fence data.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-11 23:26:19 -04:00
Digimer
597d9413a5 * Created the skeleton Cluster.pm.
* Got anvil-join-anvil to the point where is initializes and starts the cluster.
* Deleted the old ssh key handling logic in anvil-daemon.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-10 00:49:30 -04:00
Digimer
16157f7697 * Finished System->update_hosts().
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-07 16:37:27 -04:00
Digimer
de43ea3ac1 * Renamed all Validate->is_X to Validate->X. Also created Validate->ipv6() to validate IPv6 addresses using Data::Validate::IP (and added it as a requirement to the .spec base RPM).
* Added the fix from the last commit for System->call to handle returned data without an ending newline to Remote->call.
* Got more work done on System->update_hosts(). It's able to add new hosts, but misses the short and FQDN host names. Need to fix that and the verify existing / manual entries aren't molested.

Signed-off-by: digimer <digimer@pulsar.alteeve.com>
2020-07-07 01:18:38 -04:00
Digimer
76b6550ac6 * Created Database->get_ip_addresses() that pulls the IPs out and stores them in a hash that allows for easy referencing to associated interfaces and networks.
* Created Get->trusted_hosts() that finds the dashboards the host uses and, if the host is in an Anvil!, the peers in the same anvil.
* Created (but not finished yet) System->update_hosts() that will add and edit entries for all IPs to trusted hosts.
* Fixed a logging bug in Striker->load_manifest().
* Fixed a bug in System->call where, the the output from the shell call didn't end in a new-line, it would not parse the return code and lease the return code string appended to the shell output.
* Fixed a big in System->change_shell_user_password() where a new-line (\n) meant for the shell call wasn't escaped properly. There was also a duplicate 'return_code' variable preventing the actual return code from being read.
* Got more work done on anvil-join-anvil to update the hacluster password (when needed).

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-03 18:11:56 -04:00
Digimer
aa2fdfb609 * Fixed a bug in Database->get_anvils() that was clearing the manifests hash.
* got anvil-join-anvil working up to the start of configuring pacemaker.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-07-03 00:49:08 -04:00
Digimer
453f5c6223 * Fixed a bug where $anvil->nice_exit() was being passed 'exit' instead of 'exit_code' as a parameter.
* Update striker manifest run to add an entry into the 'anvils' table, and pass the anvil_uuid to the jobs rather than the various host_uuid's.
* Fixed a bug in the 'anvils' SQL procedure that copied data into the history schema (a few columns were missing).
* Updated anvil-configure-host to reboot when finished to be certain network changes have taken effect. Also updated the handling of virsh bridges to delete the autostart symlinks if libvirtd daemon isn't running.
* Added some logic to anvil-daemon to call 'anvil-update-states' with the -v{1,3} flag depending on the active debug level.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-24 00:39:56 -04:00
Digimer
1e89ef55f3 * Updated Network->get_ips() to (again) record the MAC addresses and to create a MAC to interface name lookup hash. This was (accidentally?) removed back when the ->get_ips() was changed to store the data in a host-specific hash.
* Updated tools/anvil-configure-host to use the new MAC to interface name lookup hash.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-19 12:01:29 -04:00
Digimer
bd9e74a7b1 * Added missing packages to the RPM spec and the RPMs to provide on Striker.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-18 18:46:04 -04:00
Digimer
b076463b5f * Fixed a bug in multiple System daemon control methods where the return_code wasn't being returned properly.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-18 15:17:34 -04:00
Digimer
9ab03e1f17 * Added a check to verify that connections to multiple databases are not, in fact, going to the same host.
* Added a global switch --resync-db which takes a UUID and forces that DB to be marked as needing a resync.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-17 20:11:20 -04:00
Digimer
7d907c15ff * Created the new tools/striker-purge-host tool which purges all data about a host from the database.
* Updated Database->get_hosts() to store 'host_key' and 'host_uuid' data.
* Created Database->get_ssh_keys().
* Fixed a couple bugs where Get->host_type() now returns 'striker' but tests checked for 'dashboard'.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-16 17:04:27 -04:00
Digimer
4489111a65 * Fixed a bug in Job->clear() where it was not doing it's one job right.
* Updated System->generate_state_json() where when the full host name was short, it wouldn't set the short host name properly.
* Fixed a bug in 'tools/anvil-manage-power' where the node wouldn't mark the reboot as complete. Resolves issue #11.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-11 16:08:06 -04:00
Digimer
726a4374d1 * Renamed the database table 'host_keys' to 'ssh_keys' to better represent what it stores.
* Updated 'variables' -> 'variable_source_uuid' to type 'uuid' and removed the 'not null' constraint.
* Updated Database->insert_or_update_variables() to check/update 'variables_source_table' and 'variables_source_uuid'.
* Created the 'trusts' database table which will, when done, tell anvil-daemon which users@machines to trust (setup passwordkess SSH).
* Created (but not finished) System->manage_authorized_keys() and moved the logic over to it from anvil-daemon.
* Changed the host types "dashboard" to "striker".
* Moved the following methods from 'System' to 'Get';
** System->get_host_type to Get->host_type
** System->get_bridges to Get->bridges
** System->get_free_memory to Get->free_memory
** System->get_os_type to Get->os_type
** System->get_uptime to Get->uptime
* Updated striker to include the host_uuid for the 'node1', 'node2' and (if chosen) 'dr1' when running a job manifest.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-10 18:26:50 -04:00
Digimer
e9e18f8e3b * Fixed a bug where the interface name wasn't quoted when down/up'ing an interface.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-03 23:36:25 -04:00
Digimer
530fb31478 * Updated Jobs->get_job_details() to use --job-uuid switch or, failing that, look for an incomplete on this host with the same command as the calling program.
* Got anvil-join-anvil to the point that it reworks the network configs, updates MTUs and configured NTP.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-06-03 21:52:13 -04:00
Digimer
7a247aca4e * Fixed a bug in Database->insert_or_update_bonds() where 'bond_bridge_uuid' is NULL would cause a SQL error.
* Updated Database->insert_or_update_ip_addresses() to search for an existing entry by 'ip_address_on_uuid' first, when possible.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-28 15:16:00 -04:00
Digimer
613a7f0c58 * Created the new anvil-join-anvil tool that will run on nodes and the DR host to pick up the job to join an Anvil! system.
* Finished the saving of a "run manifest" job menu. Included filtering out potential machines already in other Anvil! systems from the select box and updating the password fields to not trigger a browser to save/auto-complete the field.
* Fixed a bug in Database->get_hosts() caused by the attempt to immediately return with a 0 if it had been called before. Now a check is made in ->insert_or_update_manifests() where the recursive loop was possible.
* Updated the RPM spec to v.33 after releasing .32 after the last commit. Also added the core requirement for perl-Data-Validate-Domain.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-27 22:39:00 -04:00
Digimer
0dbb07dfb7 * Fixed a bug where fence option values with double-quotes where not being stored correctly.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-22 13:45:44 -04:00
Digimer
f0f949bcf0 * Created Striker->load_manifest() to load (and parse) install manifests.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-17 15:23:12 -04:00
Digimer
099bc1401a * Finished the menus to save a new Install Manifest and got the create page showing the existing manifests.
* Updated Database->insert_or_update_manifests() so that not passing in 'manifest_last_ran' is OK.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-14 17:52:14 -04:00
Digimer
e54aaad807 * Added MTU, NTP and DNS fields to install manifest step 2.
* Got the first BCN part of step 3 working.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-07 11:17:56 -04:00
Digimer
e66bc32693 * Added the ability to store, edit and delete UPSes
** Created Database->get_upses() and ->insert_or_update_upses().
** Created Striker->get_ups_data(). This parses the special 'ups_XXXX' strings.
* Updated Validate->is_domain() and added ->is_host_name() to use the Data::Validate::Domain module (which is now required in the core RPM).
* Started work on manifest handling.
* Sorted the language keys alphabetically.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-05-02 23:05:58 -04:00
Digimer
f71c16484a * Got the fence config confirmation screen working.
Signed-off-by: Digimer <digimer@alteeve.ca>
2020-03-12 00:25:17 -04:00
Madison Kelly
3838babf57 * OK, this time CentOS is actually supported. For reals.
** Needed to add a couple more packages to CentOS's package list.
** Changed the PXE kickstart template to create a dedicated '/boot' partition (raw partition or on RAID 1). This seems to be required now on 8.1.
** Added PXE's UEFI support to the template system (untested, but it's at least generated now).
* Filtered out 'debug' and 'verbose' options when configuring fence devices.
* Added an internet test to tools/striker-manage-install-target and skipped attempting to download packages when there's no internet. Also made loading the host OS info into a small function.
* Started creating the man pages.

Signed-off-by: Madison Kelly <mkelly@alteeve.ca>
2020-02-03 02:10:00 -05:00
Madison Kelly
27f921289f * Fixed PXE templates managing install targets to work with CentOS.
Signed-off-by: Madison Kelly <digimer@neutron.digimer.ca>
2020-02-02 17:39:56 -05:00
Madison Kelly
4787819ac9 * Restored a missing closing array bracket.
Signed-off-by: Madison Kelly <digimer@neutron.digimer.ca>
2020-02-01 21:33:37 -05:00
Madison Kelly
83d6cff4af * Fixed a bug with striker's initial configuration where 'update_value_only' was being set for the initial store of the config variables.
* Added fonts to Striker's RPM list and to the anvil-striker RPM dependency list so that the terminal is actually useful.

Signed-off-by: Madison Kelly <digimer@neutron.digimer.ca>
2020-01-23 21:45:27 -05:00
Digimer
b8c0577b54 * Fixed several issues with the fence configuration menu in striker.
* Added filters Striker->get_fence_data() for parameters. Manually change 'action' entries from 'string' to 'select' and use the data in the 'actions' element to populate it, with actions that don't make sense filtered out.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-01-22 15:21:35 -05:00
Digimer
818ef23634 * Moved the fences_unified_metadata file from /tmp, which apache can not read, to /var/www/html/.
* Fixed a bug (well, made a work-around for an issue without a known reproducer) where, on some occassion, a record will end up in the public table without being copied into the history schema. When this happens, the next resync would crash out because the resynd reads in the history table only. Now, when about to INSERT a record into the public schema during a resync, an explicit check is made to see if the record alread
y exists. If it does, the INSERT is instead redirected to the history schema.
* Cleaned up the fence agent metadata when displaying to a user, converting the shell codes to underline a string with square brackets instead. We also now replace newlines with <br /> tags. Lastly, to help fence_azure_arm's metadata description to display cleanly, a check is made to format the table correctly.
* Began work on the Striker menu for handling fence device management

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-01-20 23:41:01 -05:00
Digimer
f636e399d7 * Created tools/striker-parse-fence-agents which finds all the available fence agents on a system and gathers their metadata into a common XML file.
* Created Striker->get_fence_data() that reads/parses the unified fence metadata file created by tools/striker-parse-fence-agents.
* Created the new 'fences' database table and Database->insert_or_update_fences() to handle it.
* Added hosts -> host_ipmi that will, later, store information on how to access the host's IPMI interface, when available.
* Sketched out how the new Install Manifests are going to work.

Signed-off-by: Digimer <digimer@alteeve.ca>
2020-01-15 17:04:18 -05:00