Set bit 4 (0x10) of the requested privilege byte in RAKP Message 1
for name-only user lookup, matching ipmitool behavior. Use the same
value consistently in all HMAC calculations (RAKP2 verification,
RAKP3 auth code, SIK derivation).
Without this, some BMCs fail user lookup with "Unauthorized name"
even though the credentials are correct.
Ref: #7511
OpenBMC-based BMCs return message tag 0 in RAKP2/RAKP4 instead of
echoing the tag from the request. xCAT rejected these as stale
responses and retried indefinitely until timeout.
Accept tag 0 but verify the remote console session ID in the response
matches our current sidm. This prevents stale retries from corrupting
session state while allowing OpenBMC responses through.
Applied to got_rmcp_response, got_rakp2, and got_rakp4.
Ref: #7511
cbc_pad in decrypt mode reads the last byte as the pad count, then
calls splice(@block, 0 - $count). If decrypted data is corrupt, the
pad count can exceed the array size, crashing with "Modification of
non-creatable array value attempted, subscript -16".
Return empty string on invalid padding so the caller treats it as a
decryption failure rather than accepting corrupted data as a valid
IPMI response.
Ref: #7511
The CUDA docs were frozen at CUDA 9.2 / RHEL 7.5 / Ubuntu 14.04 since
2019. Update to cover all currently supported OS and architecture
combinations (EL 7-10, Ubuntu 20.04-24.04, x86_64/ppc64le/sbsa).
Consolidate the version-specific repo and osimage pages into generic
guides that use placeholder variables, reducing 7 files to 2 while
covering more OS versions. Both online (direct NVIDIA repo URL) and
offline (dnf download / apt download mirroring) workflows are
documented.
All NVIDIA repository URLs validated against
developer.download.nvidia.com/compute/cuda/repos/ and confirmed
accessible with valid repodata.
Addresses #7373
Previously, makedhcp warned but still created host entries without
a static IP reservation when a node's address fell inside the
dynamic range. The node would silently get a random IP from the
pool instead of its configured address.
Now errors and skips the node on all four DHCP paths (ISC v4/v6,
Kea v4/v6) with a clear message telling the admin to move the IP
outside the range or adjust the dynamic range.
This makes ISC DHCP and Kea behavior consistent and aligns with
xCAT's design: the dynamic range is for hardware discovery,
known nodes should have static IPs outside it.
Closes#6539
The quiet kernel parameter was hardcoded in anaconda.pm and sles.pm,
making it impossible for admins to get verbose boot output without
editing plugin source code. The existing addkcmdline mechanism
(bootparams and linuximage tables) only appends to the kernel command
line, so there was no way to remove quiet.
Move quiet out of the plugin kcmdline construction and into the
linuximage.addkcmdline default set during copycds osimage creation.
Admins who want verbose boot for debugging can now remove it per
osimage:
chdef -t osimage <image> addkcmdline=""
New osimages get addkcmdline="quiet" by default. Existing osimages
with a custom addkcmdline are not overwritten on re-run of copycds.
Genesis/discovery boot (mknb.pm) is unchanged as it does not use
osimage definitions.
Addresses #6916
Template.pm silently continued rendering kickstart templates when
getipaddr() failed to resolve the master hostname, producing
kickstarts with an empty MASTER_IP. Nodes would install successfully
but fail on first reboot when post.xcat and xcatinstallpost tried
to contact the master, timing out after 90 retries with:
the network between the node and is not ready
Postage.pm (mypostscript generation) already checks for this and
returns a clear error. Apply the same pattern in Template.pm so
nodeset fails immediately with a descriptive message instead of
producing a broken kickstart.
Fixes#7544
The default xCAT Apache configuration shipped with Options Indexes
enabled for the /install and /tftpboot directories. This allowed
unauthenticated users to browse directory listings, disclosing the
full tree of postscripts, boot files, and (in production deployments)
potentially kickstart files with password hashes, custom scripts with
embedded credentials, and cluster topology details.
Replace Options Indexes with -Indexes in all four shipped Apache config
files (MN and SN, Apache 2.2 and 2.4 variants). Direct file access
by known path continues to work, so all provisioning workflows are
unaffected. Directory browsing for /xcat-doc is preserved as it
contains only public documentation.
Additionally, add an Apache hardening guide documenting recommended
permissions for sensitive directories under /install, network binding
best practices, and IP-based access control options.
Addresses #7450
On some BMCs (notably Supermicro), a GET immediately after SET
returns the old value until the BMC applies the change. This made
rspconfig output misleading for network setting operations.
- Store the canonical SET value after normalization and compare
with the GET readback for ip, netmask, gateway, and backupgateway.
When they differ, annotate the output:
"BMC Gateway: 10.20.0.1 (requested 10.20.0.254, not yet reflected)"
- Consolidate ip/netmask/gateway/backupgateway display into one block
- Fix backupgateway SET: was routed through the gateway branch
writing parameter 0x0C instead of 0x0E. Now has its own branch
writing the correct IPMI parameter.
- ip=dhcp is unaffected (separate code path, never stores a value)
Tested on Supermicro IPMI BMC (10.20.0.51).
Fixes#3445
Docker container lifecycle management (mgt=docker, mkdocker, rmdocker,
lsdocker) was added in 2015-2016 as an experiment targeting Docker API
v1.22 on Ubuntu only. Documentation and man pages were deliberately
removed in 2019 (PRs #6222 and #6324) with the original developer's
approval, noting that "the interface of Docker has become very simple
right now, so there is no value for xCAT to offer such functions."
The plugin was still being shipped but has had no functional code changes
since April 2016, was never listed as a valid mgt value in Schema.pm,
and no user ever filed an issue about it.
Removed:
- xCAT-server/lib/xcat/plugins/docker.pm (1,142 lines)
- xCAT/postscripts/setupdockerhost
- xCAT-server/share/xcat/scripts/setup-dockerhost-cert.sh
- xCAT-test/autotest/testcase/dockercommand/ (test cases)
- Docker attribute definitions in Schema.pm
- Client symlinks (mkdocker, rmdocker, lsdocker)
- Usage entries and dockerhost cert handling in credentials.pm
- Docker attribute documentation in man7 pages
The "Running xCAT in Docker" documentation (dockerized_xcat/) is
retained as it documents containerizing xCAT itself, not the removed
mgt=docker feature.
Closes#7518
rspconfig vlan= only accepted values 1-4096 with no way to disable
VLAN tagging. Users had to resort to raw IPMI commands to clear a
stale VLAN after ip=dhcp.
- Accept vlan=off/disable/disabled to clear VLAN tagging via
standard IPMI parameter 0x14 with the enable bit unset
- Fix valid range from 1-4096 to 1-4094 (IEEE 802.1Q)
- Use strict digit matching to reject malformed inputs
To clear VLAN after a DHCP reset: rspconfig <node> vlan=off
Tested on Supermicro IPMI BMC (10.20.0.51).
Partially addresses #3725
Two pre-existing bugs in the alert on/off conditions:
1. Operator precedence: 'and' with 'or' without parens caused any
subcommand with argument matching /^en/ or /^dis/ to silently
trigger the alert handler.
2. Loose prefix matching: /^en/ and /^dis/ accepted typos like
"enterprise" or "discover". Replace with exact token matching
while preserving the "en"/"dis" abbreviations used by snmpmon.pm.
* Fix silent failure when site.master is not set (#6157)
Hardware control commands (rpower, rinv, etc.) silently return no output
and exit 0 when site.master is empty. The original fix (#6074) was
reverted (#6158) because it warned per-node with the wrong hostname.
Check once in plugin_command before dispatching to plugins, so the error
appears exactly once with the correct command name.
* Also reject empty site.master, not only undef
xcatd forks child processes to handle plugin requests. The child
inherits NodeRange.pm's module-level caches (@allnodeset, %allgrphash,
@grplist) with their timestamps from the parent. If the parent had
populated these caches within the past 5 seconds, the child reuses
stale data that does not reflect database changes committed by other
requests that completed between cache population and the fork.
This causes non-deterministic failures in group-definition regression
tests (chdef_group, mkdef_group, rmdef_group) where lsdef -s runs
noderange expansion inside the forked plugin process and hits the
inherited stale cache that predates the mkdef -t group commit.
Track the PID at cache-build time and treat any cache built by a
different PID as expired, forcing a fresh database read in children.
The else branch handles a rare case where XS libraries (Sys::Virt,
Net::SNMP) crash without setting $@. This IS a plugin bug, so keep
that label and the debug hint. Only the common case (die with $@)
gets the clean passthrough.
When a plugin dies during request processing, xcatd wrapped the error
in a misleading "plugin bug" message that hid the real cause (e.g.
"No space left on device"). Now passes through the actual error from
the eval, making the output useful for any failure, not just disk full.
Fixes#2719
RHEL 7 shipped ppc64le ISOs under the "alternate" label, causing
copycds to create distro paths like rhels7.6-alternate/ppc64le.
This mismatched osver() which returns rhels7.6 since /etc/os-release
has no knowledge of the alternate designation.
The architecture (ppc64le vs ppc64) already differentiates the
builds, making the alternate suffix redundant. Strip it during
auto-detection so copycds paths match osver() output.
Fixes#5593
Vinícius Ferrão
Markus Hilger