Replace eu.gcr.io/vnc-development image refs with gitea.saas.vnc.biz/vnciac,
swap gcr-json-key pull secret for gitea-registry, and update CI/deploy docs
from GitLab CI to Gitea Actions.
prosody.events.fire_event is a plain function (event_name, event_data), not
a method. The colon form prosody.events:fire_event(name, data) passed the
events table as event_name, so the lookup found no handlers and the
vnc-fcm-invalidate-notify-cache / vcard-cache signals were silently lost.
mod_vnc_muc_fcm hooks these via module:hook_global (which registers under
the string event name), so its notify_cache was never invalidated cross-host.
A stale empty cache entry for an MUC affiliate (left by an earlier test
before the user had a token) then suppressed FCM pushes to that affiliate,
breaking test_muc_fcm_push_to_offline_member in the full suite.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/11>
fcm_notify splits userName internally via jid_split, so callers were
double-splitting: passing jid_split(to) fed the bare node as userName
and shifted every subsequent argument by one (host→title, resource→body,
…), corrupting all notification payloads. Pass the full JID instead.
In mod_vnc_muc_fcm, replace event.stanza:clone() with st.clone() and
add the util.stanza import for robustness.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/11>
PostgreSQL rejects INSERT ... ON CONFLICT ... DO UPDATE on any table
that has a conditional (WHERE) DO/DO ALSO rule or a non-NOTHING DO
INSTEAD rule, erroring with 'INSERT with ON CONFLICT clause cannot be
used with table that has INSERT or UPDATE rules'. Prosody 13's
mod_storage_sql uses ON CONFLICT upserts against the prosody kv table
whenever prosody_unique_index exists (created by these scripts), so the
five inherited conditional DO rules on prosody (cache_group_avatarids,
update_profile_queue_from_insert/_update, update_muc_remote_name,
update_room_nick_jid_map_remote) broke every kv upsert (vcard, vcard_muc,
muc_remote, config, fcmtoken, ...).
Replace those rules with AFTER INSERT / AFTER INSERT OR UPDATE
row-level triggers, which do not block ON CONFLICT. The two profile-queue
rules merge into one AFTER INSERT OR UPDATE trigger so the UPDATE branch
of an ON CONFLICT upsert is also covered (it fires AFTER UPDATE triggers,
not AFTER INSERT, when the conflict is taken).
Also drop the legacy 0.11.6 update_group_owners rule. It is logically
dead under 13.0.6 (fires on key='_affiliations', which is never written)
but PostgreSQL checks rule existence at plan time, so even a dead
conditional rule blocks ON CONFLICT. Dropping it is mandatory, not
optional as the README previously claimed.
Conversion is added to both prosody-13-new-deployment.sql (fresh
deployments, and the run_new_deployment branch of migrate.sh) and
prosody-13-rules-triggers.sql (the run_rules_triggers branch for
0.11.6->13.0 upgrades), so every helm pre/post-upgrade hook path
reaches the fix. Derived-table INSTEAD upsert rules are unchanged.
Verified against postgres:16: both scripts apply cleanly, pg_rewrite
for prosody returns 0 rows, ON CONFLICT upserts succeed, and triggers
fire on both INSERT and conflict-UPDATE branches; reproduced the
production error by re-adding the legacy rules, then confirmed the
incremental script resolves it. Idempotent re-run safe.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/10>
mod_auth_http_async blocks the c2s async runner on an HTTP call inside
SASL plain_test. If the client disconnects during that call,
sessionmanager.retire_session nils every session field (incl.
base_type) and marks it destroyed. When the runner resumed,
mod_saslauth crashed at sasl_process_cdata line 94 on
'sasl/'..session.base_type..'/'. Bail out when the session is gone
instead of firing the event and sending a reply to a dead connection.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/9>
The pre-upgrade hook failed with "could not create unique index
prosody_unique_index" because the prosody table contained duplicate
rows (same host, user, store, key). Without the unique index,
mod_storage_sql falls back to SELECT-then-INSERT instead of ON
CONFLICT upsert, which races under concurrent writes and inserts
duplicates — most commonly in the fcmtoken map store.
Add a DELETE that removes duplicate rows (keeping the last-written
row per group via ctid ordering) immediately before the CREATE UNIQUE
INDEX in both prosody-13-new-deployment.sql and
prosody-13-migration-once.sql. The DELETE is a no-op when no
duplicates exist, preserving idempotency.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/8>
The mod_websocket continuation-frame patch accumulated fragments in an
unbounded dataBuffer with no size limit, fragment count limit, or
timeout. A buggy or malicious client sending endless continuation
frames without FIN could exhaust memory and degrade the entire server.
Add a configurable websocket_max_message_size option (default 2 MB)
that closes the connection with code 1009 ("Message too big") and
resets the buffer when exceeded. Also add wss-perf-analysis.md
documenting the WebSocket performance and drop investigation.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/8>
mod_vnc_broadcast and mod_vnc_fcm accessed the deprecated
prosody.hosts[host].modules.muc.rooms[to] table, removed in Prosody 13,
triggering 'Attempt to read a non-existent global rooms' warnings and
silently breaking MUC affiliation fan-out for broadcasts and push.
Replace with a version-safe get_room_from_jid() helper (rawget fallback
to .rooms for older Prosody), matching the pattern already in
mod_vcard_muc. Guard _affiliations access behind a nil room check.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/7>
Add a module-scoped util.cache LRU (lastactivity_cache_size default 5000,
lastactivity_cache_ttl default 300s) in front of the activity_store /
remote_activity_store fallback. Write-through on all set sites
(authentication-success, local last-session unavailable, remote
unavailable) and read-through in both IQ handlers (batch and standard
jabber:iq:last), so repeated last-activity queries for the same contacts
no longer hit the synchronous SQL driver, reducing event-loop blocking.
Also removes dead code (avatar-hash helpers, dumpTable, unused imports,
commented-out debug logging) and the unused remote avatar-update tracking.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/5>
Prosody 13's mod_storage_sql upgrade path only warns when the index is
missing (it is created only on fresh tables). On DBs upgraded from 0.11.6
the table pre-existed, so the index was never created and has_upsert_index
stayed false, disabling ON CONFLICT upserts and degrading write performance.
Add an idempotent CREATE UNIQUE INDEX IF NOT EXISTS to both the migration
and new-deployment scripts.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/4>
The pre-upgrade hook failed with 'permission denied for table
pg_ts_config_map' because the migration runs as prosodyDBuser (a
non-superuser), and pg_ts_config_map is a PostgreSQL system catalog that
only superusers can DELETE from.
Replace the direct 'delete from pg_ts_config_map where ...' with the
equivalent non-superuser DDL 'ALTER TEXT SEARCH CONFIGURATION ... DROP
MAPPING IF EXISTS FOR <token types>', enumerating every token type the
block re-adds. Verified against postgres:15 as a non-superuser role: the
DELETE reproduces the exact error, DROP MAPPING IF EXISTS succeeds, and
the drop+re-add round-trip is idempotent across repeated runs.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
The http (5280), telnet (5582), and web2 (8080) Services were hardcoded
as NodePort, exposing them on every node's IP. Only s2s (5222/5269)
needs external exposure and remains LoadBalancer. Switch the three
internal services to ClusterIP so they are reachable only via ingress or
within the cluster. Also update the unused service.type value in
values.yaml to ClusterIP so NOTES.txt lands in the port-forward branch
instead of the stale NodePort branch.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Switch hook-delete-policy from hook-succeeded to before-hook-creation on
both the pre-upgrade and post-upgrade migration Jobs. hook-succeeded
deleted the Job immediately on success, making post-deploy log
inspection impossible. before-hook-creation keeps the most recent run's
Job and logs around (inspectable via kubectl logs job/<name>) and only
removes the previous Job when the next deployment fires the hook,
avoiding name conflicts and accumulation.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
1. REST 404 "Unknown host: prosody": the tester reaches prosody by
service name (http://prosody:5280), so aiohttp sends Host: prosody,
which Prosody rejects as an unknown vhost. The auto host-header
heuristic only overrides for IP/localhost. Set REST_HOST_HEADER=
example.com explicitly in the tester env so HTTP routing lands on the
example.com VirtualHost that serves mod_http_rest /rest.
2. healthcheck.sh not found: the static test_04_infra checks resolve
../config/healthcheck.sh (= /config/healthcheck.sh) but the tester
image only ships /tests. Mount ./config:/config:ro so the path
resolves inside the container.
3. telnet non-loopback banner: read only 256 bytes, capturing just the
ASCII-art top and never the literal "Prosody" text. Bump to 1024 to
match test_04_infra.test_telnet_banner, which passes with the larger
read.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
The tester container uses WORKDIR /tests, so 'tests/pytest.ini' resolves
to /tests/tests/pytest.ini which does not exist, causing FileNotFoundError
in CI. Use paths relative to /tests (pytest . -c pytest.ini) for all
in-container invocations: CI job, Makefile target, tester image CMD, and
docs. Host-run scripts are unchanged (they run from repo root).
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Dockerfile: add postgresql-client and bake db-customization/ SQL scripts
into the image at /vnc/db-customization/.
config/migrate.sh: detection + migration script with two modes:
- pre-upgrade: runs idempotent prosody-13-new-deployment.sql on 13.0.x
databases; skips 0.11.6 (unsafe pre-upgrade) and new deployments.
- post-upgrade: waits for Prosody table, then runs the appropriate
scripts — full 0.11.6->13.0.6 migration (rules-triggers + migration-once
+ new-deployment) or idempotent drift correction for 13.0.x.
Helm chart: two Job templates (db-migration-pre-upgrade.yaml,
db-migration-post-upgrade.yaml) gated by dbMigration.enabled (default
true). Both reuse the Prosody image and DB credentials from existing
values. backoffLimit: 0, hook-delete-policy: hook-succeeded.
Also tracks the db-customization SQL files (previously untracked, now
referenced by the Dockerfile ADD).
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Adds a detection query set and decision tree to determine from the
database itself whether it is a new deployment, an unmigrated 0.11.6
database, an upgraded 0.11.6 database, or a clean 13.0 state — and which
script to run in each case.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Documents which SQL script to run for new deployments vs existing 0.11.6
database upgrades, ordering prerequisites, and what changed in the MUC
storage layout between 0.11.6 and 13.0.6.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
The store :set() calls go through mod_storage_sql (LuaDBI), a blocking C
client. These hooks fire on the presence/auth hot path, so inline writes
stall the event loop under presence storms. Buffer writes in memory and
flush every 5 seconds (configurable via vnc_lastactivity_flush_interval).
The IQ handlers already read from in-memory map/remote_act_cache first,
so read consistency is preserved. Final flush on server-stopping and
host-deactivating prevents data loss on graceful shutdown.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Prosody 13.0.6 removed continuation-frame (fragmented-message) support
from mod_websocket — validate_frame hard-rejects any frame with FIN=false
(close code 1003). VNCtalk clients fragment large WebSocket messages
(e.g. vCard sets with avatars >~64 KB), so the connection is silently
reset. The rejection fires on the partial-frame validation path before
handle_frame is reached, and websocket_close() does not log, so nothing
appears in prosody debug logs.
This patch removes the blanket FIN=false rejection from validate_frame
and restores the dataBuffer fragment-accumulation logic from 0.11.6 in
handle_frame: text frames (0x1) with FIN=false start a buffer,
continuation frames (0x0) append, and concatenated data is returned
only when FIN=true arrives.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Neither _hin variant is enabled in the config (the active push modules
are mod_vnc_fcm for 1:1 and mod_vnc_muc_fcm for MUC). Drop the dead
files and update AGENTS.md to reference the active modules.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Always sending Host: <xmpp domain> to /rest broke when REST was fronted
by an ingress (TLS hostname mismatch -> 431). Add a --rest-host-header
option (REST_HOST_HEADER env, default 'auto') that sends the XMPP domain
Host only for IP/localhost URLs where prosody needs it for vhost routing;
'none' disables, any other value is sent verbatim. Documents the option
in tests/README.md.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Pin the version and expected SHA256 as build ARGs and verify the
download with sha256sum -c so a tampered or corrupt source download
fails the build loudly. Also drops the noisy tar -v and removes the
tarball after extraction.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Bumps the base image to docker:24.0 (needed for Usage: docker compose [OPTIONS] COMMAND
Define and run multi-container applications with Docker
Options:
--all-resources Include all resources, even those not
used by services
--ansi string Control when to print ANSI control
characters ("never"|"always"|"auto")
(default "auto")
--compatibility Run compose in backward compatibility mode
--dry-run Execute command in dry run mode
--env-file stringArray Specify an alternate environment file
-f, --file stringArray Compose configuration files
--parallel int Control max parallelism, -1 for
unlimited (default -1)
--profile stringArray Specify a profile to enable
--progress string Set type of progress output (auto,
tty, plain, json, quiet)
--project-directory string Specify an alternate working directory
(default: the path of the, first
specified, Compose file)
-p, --project-name string Project name
Management Commands:
bridge Convert compose files into another model
Commands:
attach Attach local standard input, output, and error streams to a service's running container
build Build or rebuild services
commit Create a new image from a service container's changes
config Parse, resolve and render compose file in canonical format
cp Copy files/folders between a service container and the local filesystem
create Creates containers for a service
down Stop and remove containers, networks
events Receive real time events from containers
exec Execute a command in a running container
export Export a service container's filesystem as a tar archive
images List images used by the created containers
kill Force stop service containers
logs View output from containers
ls List running compose projects
pause Pause services
port Print the public port for a port binding
ps List containers
publish Publish compose application
pull Pull service images
push Push service images
restart Restart service containers
rm Removes stopped service containers
run Run a one-off command on a service
scale Scale services
start Start services
stats Display a live stream of container(s) resource usage statistics
stop Stop services
top Display the running processes
unpause Unpause services
up Create and start containers
version Show the Docker Compose version information
volumes List volumes
wait Block until containers of all (or specified) services stop.
watch Watch build context for service and rebuild/refresh containers when files are updated
Run 'docker compose COMMAND --help' for more information on a command. v2) and
adds a test:compose job that builds the stack, waits for prosody health,
runs the slixmpp suite in the tester container, and emits a junit report.
Runs on merge_request_event and non-main branch pipelines.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
- chart: add secretEnv (existingSecret + keys) rendered as secretKeyRef in
both containers; add resourcesProsody/resourcesSidecar defaults with
dedicated requests and limits via a containerResources helper
- argo: move prosodyDBpass/fcmApiKey/fileShareSecret/avatarUploadPass out of
env into secretEnv referencing a prosody-secrets Secret; set dedicated
resource requests+limits per container
- argo: add prosody-secrets.example.yaml ArgoCD Application (with a warning
to use Sealed Secrets / External Secrets / SOPS instead of plaintext)
- chart fixes: bump stale appVersion 0.11.6 -> 13.0.6; fix malformed service
block in values.yaml
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Set PROSODY_DEBUG=true to switch the prosody.log level from info to
debug. This makes stanza-too-large rejections and XML parse errors
visible — both are logged at debug level and otherwise invisible.
Usage in k8s:
env:
- name: PROSODY_DEBUG
value: "true"
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Add c2s_stanza_size_limit and s2s_stanza_size_limit to the config
template, backed by C2S_STANZA_SIZE_LIMIT and S2S_STANZA_SIZE_LIMIT
env vars. Both default to 5242880 (5MB) when unset, set in startup.sh.
The compose harness sets them explicitly.
Prosody 13.0.6 defaults are 256KB (c2s) and 512KB (s2s), which are too
small for large file-transfer invitations and Jitsi meet sessions.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Combined results from compose harness (80 passed, 6 skipped) and external
deployment (50 passed, 36 skipped). 81 unique tests passed across both
runs, 0 failures. 5 remaining skips are config/infrastructure prerequisites,
not regressions. All 81 passed tests described with what they verify.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
The test was skipping because it depended on the second_client fixture,
which uses XMPP_JID2 (a distinct account). Carbons require both
resources to share the same bare JID.
Rewrite to create a second VNCXmppClient inline with the same bare JID
as xmpp_client but a different resource (/carbon-<random>), so the test
runs regardless of whether XMPP_JID2 is configured or distinct.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Check if the SSH tunnel to PostgreSQL (port 14322) and the kubectl
port-forward for telnet (port 5582) are active before running the
testsuite. Start them automatically if missing. Clean up the
kubectl port-forward on exit if we started it.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Bump Dockerfile to prosody-13.0.6; all 8 patches re-ported against 13.0.6.
Re-ported patches (3 changed, 5 applied with offset):
- hidden.lib.patch: 13.0 uses module:may() instead of um_is_admin; changed
to 'if restrict_public then' (same intent: hide option for everyone)
- mod_muc.patch: 13.0 added restrict_pm between register and
presence_broadcast; updated hunk 1 context
- mod_muc_unique.patch: 13.0 uses 'require "prosody.util.stanza"'
(namespaced); updated context
- muc.lib, mod_carbons, mod_mam, mod_muc_mam, register.lib: applied
with line offsets, no re-port needed
Config changes:
- Remove mod_posix from modules_enabled (13.0 absorbed signal handling,
pidfile, and run_as_root check into core util/startup.lua)
- Move pubsub from modules_enabled to Component (13.0 requires pubsub
to be loaded as a component, not a module)
Test fix:
- test_muc_fcm_push_to_offline_member: wait for count=2 captures instead
of 1 — mod_vnc_muc_fcm pushes to ALL affiliated members (including
sender, because it can't see main-host sessions from the MUC
component); the test was racing on which push arrived first
Verified: prosodyctl check config passes; compose-harness testsuite
green (80 passed, 6 skipped, 0 failed).
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Update M2 status to reflect both compose-harness (80 passed, 6 skipped,
0 failed) and external testsuite results (48 passed, 30 skipped — all
skips are infrastructure prerequisites). test_vcard_fallback now skipped
pending mod_vnc_vcard_fallback enablement.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
The module exists in vnctalk/ but is not listed in modules_enabled in
the config template. The test previously passed on external deployments
by coincidence — the test users already had vCards with FN from real
usage. On the compose harness with a fresh DB, the test fails because
nothing generates a vCard.
Add @pytest.mark.skip with a reason pointing to the missing module.
Simplify run-tests.sh to a single pytest invocation (was 12 separate
calls). Update m1-manual-tasks.md §3.5 to reflect the skip.
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
Brings up the docker-compose stack (postgres + mocks + prosody), waits
for prosody to become healthy, and runs the full pytest suite with all
env vars set so tests that need MOCK_URL, PG, BOSH/WS, telnet, etc. are
not skipped.
Auto-detects if port 5582 is already in use (e.g. kubectl port-forward)
and remaps the telnet host port to 5583 via the TELNET_HOST_PORT env var
in docker-compose.yml.
Usage:
./run-compose-tests.sh # up + test + down
./run-compose-tests.sh --no-down # leave stack running
./run-compose-tests.sh --down # just tear down
Part-of: <http://gitlab.vnc.biz/uxf/vnctalk-prosody/-/merge_requests/3>
macanhhuy
Stefan-Sanger