mirror of
https://github.com/Cacti/cacti.git
synced 2026-10-25 04:27:56 +09:00
develop
4
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 TheWitness
|
f79dc15c08 |
Port #8166 SNMP credential cache + procedural SNMPv3 (with #8171/#8172 follow-ups) to develop (#8174)
* Port #8166 SNMP credential cache + procedural SNMPv3 to develop Ports the shared SNMPv3 credential cache and procedural SNMPv3 routing from the 1.2.x feature (merged as #8166) onto develop, adapted to develop's refactored SNMP layer. Parts of the 1.2.x change that do not apply to develop are omitted: the cacti_exec() idle-timeout rewrite (develop has no cacti_exec()), the PHP floor bump and composer/doc changes (develop is already 8.3+), and the cmd.php max_oids batching (develop's cmd.php already collects SNMP in-process through persistent sessions, so the per-item spawn the batching removed does not occur). - lib/cache.php: new cross-process \Cacti\Cache\SharedCache (file + POSIX shared-memory + process-local tiers, optional AES-256-GCM sealing). Loaded via lib/snmp.php (the Cacti\Cache\ namespace is not PSR-4 autoloaded). The crypto helpers it seals with already live in develop's lib/auth.php, loaded by global.php, so no duplicate was added. - lib/snmp.php: snmp_get_method() now routes SNMPv3 get/getnext to the procedural snmp3_*() calls when the running PHP supports the configured auth/privacy tokens and no context/engine id is set (SHA-224/384 and AES-192/256[C] stay on the binary before PHP 8.6); the three single-get functions gain the procedural v3 path with \Throwable handling; adds the credential-cache subsystem, protocol-option/MD5-DES helpers, snmp_native_protocol(), snmp_format_agent() and cacti_snmp_get_multi(); the binary v3 auth builder consults the cache when enabled. - poller.php / script_server.php: prime / load the shared cache. - poller_maintenance.php: daily out-of-band cache rebuild on every collector (per-poller lastrun marker). - lib/api_device.php / lib/utility.php: bump snmp_cred_version only on v3 credential add/change (and propagate the token to the affected remote poller). - include/global.php: resolve and create the private cache/ directory. - include/global_settings.php: add 'Enable Credential Cache' and 'Enable MD5 and DES for SNMPv3' settings (both off by default); default new-device v3 auth/privacy to SHA/AES instead of legacy MD5/DES. - host.php / automation_snmp.php: drop MD5/DES from the pickers when disabled and default new records to SHA/AES on upgraded installs. - include/config.php.dist: document $config['cache_dir']. - tests/Unit/Core/Cache/SharedCacheTest.php: SharedCache behaviour coverage. - tests/security/baselines: baseline the two new lib/cache.php sinks. * Incorporate #8171: binary fallback for unsupported SNMPv3 protocols Layers PR #8171 onto the develop port so an SNMPv3 device whose auth/privacy protocols the running PHP's ext-snmp cannot service natively (e.g. AES-256 before PHP 8.6) no longer fatals the poller or device-edit screen. - lib/snmp.php: add CactiSnmpBinarySession, a session stand-in exposing the get/getnext/walk/close interface the cacti_snmp_session_*() helpers use and delegating each call to cacti_snmp_*() -> the Net-SNMP binary. cacti_snmp_session() returns it when php-snmp is absent or the v3 protocol is not natively serviceable, so cmd.php/host.php collect these devices via the binary on pre-8.6 and switch to the in-process path on 8.6 with no caller change. v1/v2 snmpget/snmp2_get now go through snmp_format_agent(). The ERRNO_TIMEOUT check tolerates php-snmp being absent. Drop the vendored classSNMP.php compatibility shim and its include (and the Yandex attribution). - cmd.php: open_snmp_session() returns the down sentinel when a host is already marked down this run, instead of returning an unset (null) [] that cacti_snmp_session_get() then dereferenced (TypeError). - lib/ping.php: ping_snmp() runs the availability check through the procedural cacti_snmp_get()/getnext() path so an unsupported protocol falls back to the binary instead of raising a ValueError in SNMP::setSecurity(). * Incorporate #8172: retire include/vendor/phpsnmp, relocate into lib Follow-up to #8171 on the develop port. After classSNMP.php was removed, the include/vendor/phpsnmp/ directory held only Cacti-owned files; this retires it. - lib/snmp.php: the ext-snmp subclass that carried Cacti's bulk_walk_size / value_output_format (previously vendor/phpsnmp/extension.php) is now declared directly as CactiSnmpNativeSession, guarded by CACTI_PHP_SNMP; the include and the 'use phpsnmp\SNMP' alias are gone, and new SNMP(...) / SNMP::VERSION_* now reference the native \SNMP and the new subclass. - lib/mib_parser.php: moved from include/vendor/phpsnmp/mib_parser.php; lib/mib_cache.php includes it via __DIR__. - Remove include/vendor/phpsnmp/extension.php and index.php (directory retired). - tests: SnmpCoverageTest uses \SNMP::ERRNO_TIMEOUT now that phpsnmp\SNMP is gone; drop the stale include/vendor/phpsnmp/ entry from the vendor-policy allowlist. develop's sink baseline tracks constant-based includes differently than 1.2.x, so no baseline refresh was needed here. * Address PR #8174 review: snmp_get_method params, v3 walk, getnext agent, cache bump, CHANGELOG - lib/snmp.php: restore $php_snmp as the 6th snmp_get_method() parameter and move the new $auth_proto/$priv_proto after it (callers pass them by name), so the positional $php_snmp used by SnmpCoverageTest and others keeps working. - lib/snmp.php: keep every SNMPv3 operation except get/getnext on the Net-SNMP binary - the native walk branch implements only v1/v2, so a v3 walk reaching the procedural gate would have silently returned empty when path_snmpbulkwalk is missing. - lib/snmp.php: cacti_snmp_getnext() v1/v2 branches now use snmp_format_agent() like get/get_raw, so an IPv6 literal / non-default port is bracketed. - lib/api_device.php: bump snmp_cred_version only when the host's v3 membership or one of the cached credential fields actually changed, instead of on every save of a v3 device. - CHANGELOG: move the #8171 entry into the issue block and the #8166 entries into the feature block, matching the established section ordering. * Fix CI: unbracket IPv4/DNS SNMP targets, inline auth-cache SQL columns, apply phpcsfixer standard * Refresh sink inventory baseline after short-array normalization * Restore lib/snmp.php 100% coverage after aligning v3 routing to 1.2.x - update the three SnmpCoverageTest assertions that still encoded develop's pre-port behavior: v3 snmp_get_method now returns PHP, an unsupported- protocol v3 session returns the binary-delegating session object, and the v3 cacti_snmp_get path runs natively - add a v3 noAuthNoPriv user to the snmp-coverage snmpd fixture so native snmp3_get is exercised deterministically - add tests covering the ported subsystems to 100%: shared auth cache, binary-delegating session, protocol pickers, v3 credential args, cacti_snmp_get_multi, and the cacti_get_snmpv3_auth cache path - snmp_auth_cache_load uses a resettable process guard so its enabled and disabled branches are both coverable in a single process * Resolve phpstan errors in the relocated SNMP MIB parser and ported snmp.php - lib/mib_parser.php: type the MibParser properties, add parameter/return types, guard file_get_contents()/preg_split() against false, correct the parse_bracket_token delimiter params (string, not integer), and drop the unreachable return after trigger_error(E_USER_ERROR) - now phpstan clean - lib/mib_cache.php: drop the redundant isset($mp->oids) guard (the typed property is always initialized; $mp->mib is the real check) - lib/snmp.php: type the CactiSnmpBinarySession and CactiSnmpNativeSession properties/methods and pass bool to snmp_set_quick_print() - .phpstan.neon: ignore greaterOrEqual.alwaysFalse alongside the existing greater.alwaysFalse (the PHP_VERSION_ID >= 80600 forward-compat gate) - regenerate phpstan-baseline.neon for the irreducible ext-snmp stub strictness (snmp3_get literal-union protocols, int<-1,max> timeouts) in the ported procedural v3 path, the new lib/cache.php shmop stubs, and the $config array-shape offsets in include/global.php * Make the v3 getter catch coverage net-snmp-build-independent The DES-rejection trick that drove the procedural snmp3_get()/snmp3_getnext() try/catch blocks only works on net-snmp builds compiled without DES; CI's Debian net-snmp accepts DES, so those catches went uncovered (98.3%). - extend the existing native_get injection seam (already on cacti_snmp_get) to cacti_snmp_get_raw, cacti_snmp_getnext, and cacti_snmp_get_multi, and drive the catch blocks with a throwing native getter in the test - mark the seam-less cacti_snmp_session() setSecurity catch with @codeCoverageIgnore (it only throws for a protocol the running net-snmp rejects, which is not portably reproducible in the fixture) * Exclude ext-snmp/PHP-8.6-gated defensive branches from snmp coverage Two branches in lib/snmp.php are unreachable on the PHP 8.3 + ext-snmp coverage runner and were only "covered" locally via opcache dead-code elimination: - snmp_get_method()'s !function_exists('snmp3_get') fallback (ext-snmp always provides snmp3_get) - snmp_php_v3_protocols_supported()'s PHP_VERSION_ID >= 80600 block (SHA-224/384 and AES-192/256 tokens are only accepted on PHP 8.6+) Mark both with @codeCoverageIgnore so the 100% gate is deterministic across runners regardless of opcache settings. * Account for the #8166 SNMPv3 credential-cache gate query in the batching test poller_update_poller_cache_from_buffer() now runs one SELECT per flush to decide whether the flush touched SNMPv3 poller items before bumping the shared credential-cache version (matching 1.2.x). UpdateDataQueryCacheBatchingIntegrationTest counted 4 statements per flush; update it to 5 (and 3 per-row flushes to 15, 1 batched flush to 5). The guard's intent - batching issues fewer statements than per-row flushes - still holds. * Port #8175: group SNMP options under an SNMP Behavior section Mirror the 1.2.x settings reorganization (#8175) on develop. develop already keeps the SNMP options in a dedicated SNMP tab, so add an "SNMP Behavior" spacer there, order the behavior options as md5/des, credential cache, max get size, bulkwalk fetch size, and move "Disable increasing OID Check" out of the Poller tab to sit last in that group. Display-only; no settings added or removed and no behavior change. Adds the "SNMP Behavior" heading to cacti.pot. * fix: complete dependabot htmx 2.0.10 to 2.0.11 bump Dependabot bumped package.json and package-lock.json to htmx.org 2.0.11 (PR 8179) but left the vendored include/js/htmx.js, its version marker, and the pinned SRI constant at 2.0.10. The htmx-filters CI workflow runs npm ci, which regenerates include/js/htmx.js from the 2.0.11 package, so HtmxTest's runtime sha384 no longer matches the HTMX_2_0_10_SRI constant and the integrity assertion fails. Complete the bump: regenerate the vendored include/js/htmx.js to the 2.0.11 build, bump include/js/htmx.js.version to 2.0.11, rename HTMX_2_0_10_SRI to HTMX_2_0_11_SRI with the recomputed sha384 hash, and update the version/integrity assertions in tests/Unit/Ui/Html/HtmxTest.php and tests/HandOff/HtmxLoaderHandOffTest.php. package.json and package-lock.json match the values dependabot already landed on develop. * settings: move SNMP Agent Support under the SNMP Behavior spacer Relocate the enable_snmp_agent setting (SNMP Agent Support Enabled) so it is the first item under the SNMP Behavior spacer instead of sitting in the Poller General group, grouping it with the other SNMP behavior options. * Address PR #8174 review: keep SHA-224/384 on the binary, honor port on transport targets Resolves the two still-valid Copilot review findings on lib/snmp.php that were left unresolved (only marked outdated after later commits shifted their lines). - snmp_php_v3_protocols_supported(): drop SHA224/SHA384 from the native auth allowlist. ext-snmp's SNMP::setSecurity()/snmp3_*() only map MD5/SHA/SHA256/ SHA512 - there is no SHA-224/SHA-384 mapping on any build, including PHP 8.6 (verified against php-src ext/snmp/snmp.c), so routing those tokens to the procedural path raised a ValueError instead of using the binary. They now always fall back to the Net-SNMP binary. The AES-192/256 privacy tokens that ext-snmp does add on 8.6 stay gated behind PHP_VERSION_ID >= 80600. - snmp_format_agent(): a udp/tcp transport-qualified target that omits a port now has the configured non-default port appended (udp6:[2001:db8::1] + 1161 -> udp6:[2001:db8::1]:1161), matching the Net-SNMP binary path. Targets that already carry a port, and unix: sockets (no port), are left untouched. - SnmpCoverageTest: cover the new transport-target branches; lib/snmp.php stays at 100%. |
||
|
Thomas Vincent
|
95060c49ba | Close test-suite discovery and coverage gaps (#7970) | ||
|
TheWitness
|
096b002da5 | ci: Disable two tests until fixed (#7587) | ||
|
Thomas VincentandTheWitness
|
db39178735 |
fix: harden the poller_item cache and Boost RRD staging subsystems (#7538)
* fix(boost): forward still-open-round archive rows instead of dropping them boost_process_poller_output() correctly excluded rows from an in-progress poll round when building the RRD write, but the archive-table cleanup right after deleted every row for the local_data_id anyway, with no time filter. Those rows now get INSERT IGNORE'd into the live poller_output_boost table before the archive table is cleared, so a still-open round's data survives. Also fixes the duplicate-record guard in the same function: $last_item was initialized once before the results loop and never reassigned, so the comparison against it could never be true. It is now updated after each processed row. Fixes #7519 Fixes #7523 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(boost): gate archive-table drop on shard completeness poller_boost.php decided whether to drop a run's archive tables using SELECT SUM(status) FROM poller_output_boost_processes, an aggregate over whatever completion rows landed. If one child crashed before writing its completion row, the sum from the surviving shards was still positive and every archive table for the run got dropped, including the crashed shard's unprocessed data. The drain loop already computes boost_completed_children() vs $expected_children to detect a crashed child; the drop now uses that same completeness check instead, and leaves the archive tables in place with a warning when it is not met. Fixes #7520 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(dsstats,rrdcheck): apply Boost's startup barrier to launch loops dsstats_launch_children()/rrdcheck_launch_children() fork children via exec_background() (non-blocking) and the callers then wait with a bare while ($running = ..._processes_running($type)) { sleep(2); }. If the parent evaluates that before any child has registered in the processes table, $running reads 0 and the loop body never runs, treating "not started yet" as "already done" -- the same race Boost's boost_all_children_registered() barrier was added to close. dsstats_launch_children()/rrdcheck_launch_children() now return the launched count, and dsstats_wait_for_children()/ rrdcheck_wait_for_children() wait for the running count to reach it (bounded by a deadline) before trusting the drain loop's running-count poll. Fixes #7526 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * refactor(poller): consolidate poller_item deletes behind a shared helper Extract poller_item_delete_for_data_source() and poller_item_delete_for_host() in lib/poller.php and migrate the existing local+remote delete pairs in lib/api_device.php and lib/api_data_source.php onto them, so the local-delete-plus-remote-mirror pattern lives in one place instead of eight. Refs #7535 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(poller): mirror poller_item deletes to the remote poller on handoff api_data_source_change_host(), data_query_remove_disabled_items(), and data_source_to_data_template() each moved or removed a data source's poller_item row locally but never on the remote Data Collector that still owned it, so that Collector kept polling an item it no longer should. All three now resolve the owning remote poller before the local change and purge the row there too via the shared helper. Fixes #7521 Fixes #7524 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(poller): correct poller_item join semantics and cron reset in cmd.php poller_item and host share ~13 identically-named snmp_*/hostname columns; SELECT * across the join let PDO FETCH_ASSOC silently collapse them to host's live values instead of poller_item's cached copies. Select pi.* plus explicitly aliased host/site disabled columns instead. The now already-fetched $poller_items also makes the separate script-server-call COUNT(*) re-join redundant, replaced with an in-PHP count. Also apply the same cron_interval-aware rrd_next_step reset used by the ranged branch to the --allhost branch, which previously ignored it. Fixes #7522, #7525, #7528. Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(poller): thread known poller_id through update_poller_cache() callers update_poller_cache() re-derived poller_id via a host/data_local JOIN on every call, even when the caller already had it: repopulate_poller_cache() and push_out_data_input_method() both already select h.poller_id and discarded it, push_out_host() has the host row cached per data source, and api_device_enable_devices()/update_poller_cache_from_query() compute it once outside their loops. Add an optional $poller_id argument (and honor an already-present 'poller_id' array key) so these callers skip the redundant query. Fixes #7527. Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(poller): scope global aggregate scans and make process-leveling sargable Three poller.php queries (total_snmp_ports and the mutually-exclusive active_profiles pair) had no poller_id filter, unlike the rest of the file's central-only blocks, so every remote poller redundantly recomputed and overwrote the same global config options; the snmp_port scan is also an unindexed full table scan. Gate them behind poller_id == 1, with remote pollers reading back the value the central poller already stored. Also rewrite the process-leveling $sql_where's rrd_next_step predicate from the non-sargable `rrd_next_step - $poller_interval <= 0` to the algebraically equivalent `rrd_next_step <= $poller_interval`, restoring use of the poller_id_rrd_next_step index. Fixes #7529, #7530. Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(poller): batch update_data_query_cache()'s poller cache writes Each changed data source called update_poller_cache($data_source, true), which flushes immediately via poller_update_poller_cache_from_buffer() -- its own present=0 UPDATE, INSERT ON DUPLICATE KEY UPDATE, and stale-row DELETE -- on every iteration. Re-indexing a host where every port changed issued up to 3x as many statements as ports instead of one batched flush. Accumulate changed rows across the loop and call update_poller_cache() with commit=false, then flush once via poller_update_poller_cache_from_buffer() after the loop, matching the batched pattern already used by push_out_data_input_method(). Fixes #7531. Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(boost): add USING BTREE to poller_output_boost_local_data_ids indexes MEMORY-engine indexes default to HASH unless USING BTREE is specified, unlike the sibling poller_output/poller_output_boost tables in the same schema file. boost_output_rrd_data()'s ORDER BY/LIMIT query and boost_process_local_data_ids()'s ranged DELETE both need to scan local_data_id in order, which a HASH index can't serve, forcing a filesort on every boost worker pass. Verified against a real MariaDB 11 instance: EXPLAIN on the ORDER BY query shows "Using filesort" with the old HASH schema and no filesort with USING BTREE, using the primary key for ordering instead. Also aligns boost_prepare_process_table()'s dead CREATE TABLE IF NOT EXISTS fallback with the corrected schema, and adds an upgrade step that rebuilds both indexes as BTREE on existing installs (checked via information_schema.STATISTICS, since db_install_add_key() only compares index columns, not algorithm, and would otherwise skip an existing HASH index with the right column list). Closes #7532 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(boost): skip the UNION ALL wrapper for the single-archive-table case boost_process_local_data_ids() always wrapped its flush query in "SELECT * FROM (... UNION ALL ...) t ORDER BY ...", even when there was only one archive table -- the steady-state case, since old archive tables are dropped at the end of each cycle. UNION ALL results wrapped in a derived table are one of the documented cases MySQL/MariaDB's derived-table merging doesn't apply to, so every pass paid for a filesort over an indexless derived table regardless of how well-ordered the underlying archive table's rows already were. Extracts the per-table SELECT into boost_archive_select_sql() and uses it directly, without the wrapper, when there's exactly one archive table. The UNION ALL/derived-table shape is kept as the fallback for the genuinely rare case where a prior crashed run left more than one archive table behind. Verified against a real MariaDB 11 instance and via a SQLite-backed integration test that both shapes return correctly filtered and sorted rows. Closes #7533 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(boost): batch-prefetch RRD field name lookups instead of per-row queries Two N+1 patterns in the Boost hot paths: boost_process_local_data_ids() (poller_boost.php) re-ran the unused_data_source_names/nt_rrd_field_names lookups on every local_data_id boundary in the sorted flush result set. process_poller_output() (lib/poller.php) had the equivalent lookup, but without even a boundary check -- it re-ran per matching row whenever the output contained a ':', including consecutive rows for the same data source. Adds poller_prefetch_rrd_field_names() (lib/poller.php, shared by both callers since poller_boost.php already requires lib/poller.php) to batch-fetch these mappings once per invocation via IN()-batched queries keyed by the full set of local_data_ids in the batch, chunked at 1000 ids to match the array_chunk(..., 1000) convention used elsewhere (lib/api_data_source.php, lib/api_graph.php). Both call sites now look up the prefetched maps in-memory inside their loops. While consolidating the lookups, fixes a latent bug the old process_poller_output() code had in its non-templated fallback branch: a query referencing gti.task_item_id without joining graph_templates_item, which throws an unknown-column error whenever that path is hit. The batched query always joins correctly. Integration test proves both correctness (against a real SQLite-backed schema) and the query-count reduction: 5 local_data_ids that would previously cost up to 10 queries for these two lookups now cost exactly 3, regardless of how many ids are in the batch. Closes #7534 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * refactor(boost): consolidate poller_output_boost inserts into one helper The INSERT INTO poller_output_boost (...) VALUES ... statement was hand-written four times: three in cmd.php (behind the same boost_redirect == 'on' && boost_rrd_update_enable == 'on' guard) using INSERT IGNORE, and once in boost_poller_on_demand() (lib/boost.php) using ON DUPLICATE KEY UPDATE. Beyond the duplication, that was a real data-integrity inconsistency between two paths writing the same shape of data: silently-drop vs last-write-wins on the same key. Since boost_redirect gates cmd.php's writes and boost_poller_on_demand() returns early when boost_redirect is on, the two paths never run concurrently for the same install; they're alternate write paths selected by that one setting, not competing writers. A genuine collision on (local_data_id, rrd_name, time) in either path represents a retry of an already-collected sample -- a batch resent after a partial failure, or two writers racing on the same rounded second -- not two independently meaningful readings, since RRD only stores one value per timestamp regardless. Standardizes on INSERT IGNORE (first write wins), matching the sibling poller_output insert this table shadows, which has always used INSERT IGNORE, and 3 of the 4 existing call sites. Extracts boost_flush_output_batch() (lib/boost.php) as the one shared helper, handling both the batched INSERT and the max_allowed_packet-safe chunking cmd.php's three sites previously didn't do at all (they only chunked by item count). Integration test exercises a genuine duplicate-key collision against a real (SQLite-backed) table and asserts the chosen first-write-wins behavior. Closes #7536 Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(tests): resolve cross-branch test interactions found by full-suite verification Two test-only bugs surfaced only when all four fix branches for the poller_item/Boost hardening pass ran together, not when each ran in isolation: - BoostArchiveHandoffIntegrationTest extracted the archive-forward INSERT IGNORE by searching lib/boost.php for the first occurrence of "INSERT IGNORE INTO poller_output_boost". The Boost SQL-optimization branch adds a second, earlier occurrence (the new boost_flush_output_batch() helper's single-line prefix), so the naive search now grabbed the wrong snippet. Anchor on the multi-line "...poller_output_boost\n\tSELECT *" shape that only the archive-forward statement has. - PollerItemRemotePurgeHandOffTest calls push_out_host() (via api_data_source_change_host()), which needs CACTI_PATH_LIBRARY. The test never requires include/global.php (which normally defines it), and was relying on some other file in a full multi-thousand-file suite run having defined it first as a process-wide side effect. Define it directly instead of depending on unguaranteed load order. No production code changed. Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * style(poller): satisfy php-cs-fixer strict_param in the script-server count Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(boost): order the single-archive-table flush by the table's PK columns Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(boost): key the max_allowed_packet cache per connection Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> * fix(tests): restore the default db_* session after swapping it for a fake handle Left swapped, the fake handle answered every later read_config_option() in the run; on PHP 8.4 the extra deprecation traffic routed through CactiErrorHandler turned that into an uncaught PDOException and killed the pest process. Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> --------- Signed-off-by: Thomas Vincent <thomasvincent@gmail.com> Co-authored-by: TheWitness <thewitness@cacti.net> |