[Rushaway]

Summary

• The color list grew on every reload until it overflowed. OnSQLSelect_Color appended the sb_srvgroups rows to g_sColorList without resetting g_iColorListSize and without checking COLOR_LIST_MAX_LENGTH (255). Each sm_admins_config_mod change to 0 or 1 calls SQLInitialize() and adds the whole table again. That duplicates the entries scanned for every sm_admins, and after enough changes (or with more than 255 groups) it raises an "Array index out-of-bounds" error in the callback, so the colors stop loading. Now the callback resets the size first and stops at 255 rows. (AdvancedAdminList.sp)
• Connection handling. SQLInitialize() deleted and reopened the connection on every cvar change. A change made while a connect was still in flight produced two callbacks, and both assigned g_hDatabase, so the first handle leaked. Now the open connection is reused (only the colors are reloaded), and a second call while a connect is pending does nothing. When the driver is not MySQL the handle is now closed instead of being kept, unused, as g_hDatabase. A failed connect is now logged and retried after 30 s; before, colors never loaded until the plugin was reloaded. (AdvancedAdminList.sp)
• Version 2.1.9 -> 2.1.10.

Not included

• A missing sourcebans entry in databases.cfg still calls SetFailState, as before (it is a config error, not a failed connect).

Test plan

• Compiles with spcomp 1.12.0.7259 (same deps as CI), no new warnings
• CI build passes
• Toggle sm_admins_config_mod 1 -> 2 -> 1 several times: sm_admins shows the same colors each time, and no array-index errors appear in the logs
• Start with MySQL down: a retry is logged every 30 s, and the colors load once MySQL is back

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota.

[WareZ57]

Checked this against master (merges cleanly, CI green) and compiled with spcomp 1.12.0.7259 + MultiColors, no warnings.

• OnSQLSelect_Color now resets g_iColorListSize and stops at COLOR_LIST_MAX_LENGTH (AdvancedAdminList.sp:229-235). That fixes both the duplicate entries on every sm_admins_config_mod change and the out-of-bounds write with more than 255 groups. Callbacks run one at a time on the main thread, so two overlapping SELECTs (toggling the cvar quickly) just rebuild the same list twice.
• Connection handling: an existing handle is reused (:152-156), a second call while a connect is in flight is a no-op (:159-160), and the non-MySQL handle is now deleted instead of kept (:186-191). I didn't find any path that can assign g_hDatabase twice or leak a handle now.
• The retry timer is guarded by g_hSQLRetryTimer so it can't stack. It has no TIMER_FLAG_NO_MAPCHANGE, so it survives a map change, and it re-checks the config mode before reconnecting (:198-206). Looks right.

Unrelated to this PR and optional: switching sm_admins_config_mod to 2 (".cfg only") doesn't clear g_iColorListSize, so the SQL colors from an earlier mode 0/1 are still used as the fallback in the lookup at :534-544. Resetting g_iColorListSize = 0 in OnCvarChanged when the mode becomes 2 would make it match the cvar description. That can be a separate change.

LGTM.