Update made NPCs drop 100%

[thepoetwarrior]

Some how when I updated source code and database from May 25th to current June 15th, all the NPCs in all zones got affected to have 100% drop rate. Using Georges NPC LOOT EDITOR I verified the drop % are correct, but some how server is treating them all as if they were 100%.

For now, I used my backups, and restored the tables updated by the sqls and used a backup, and put in the old binaries again, opt codes, etc.

Here were the changes made, and I don't know where / why every mob became 100% drop rates.

USED THESE SQL UPDATES
----------------------
1510_global_instances.sql
1511_map_path_loading.sql
1513_zone_points.sql
1519_zone_primary_key_id.sql
1542_items_table_cleanup.sql

AFFECTED TABLES
---------------
items
zone
zone_points
doors
instance_lockout


EQEMu Changelog (Started on Sept 24, 2003 15:50)
-------------------------------------------------------
==06/15/2010==
Secrets: Characters with HPs higher than 30k will no longer be set back to 30k current health when zoning.
gaeorn: fixed integer transformation for RNG as it previously would only return the high value of the low to high range if the RNG returned exactly 0xffffffff
Trevius: The following item fields now load from the database: EliteMaterial LDoNSellBackRate ScriptFileID ExpendableArrow QuestItemFlag.
Trevius: Identified some unknown fields in the database from 13th floor findings, and removed multiple unused fields.
Trevius: The summonedflag field is copied from UNK109 field.

REQUIRED SQL:
utils/sql/svn/1542_items_table_cleanup.sql

==06/10/2010==
gaeorn: modified method of transforming RNG output into a range of integers. this should improve the equalization of the distribution of the returned random numbers

==06/09/2010==
Trevius: Added new Perl Mob quest object SpellEffect(effect, [duration, finish_delay, zone_wide, unk20, unk26]) for sending a spell paticle effects without casting a spell.
gaeorn: replaced random number generator with the first one on this page: http://www3.ocn.ne.jp/~harase/megenerators.html

==06/08/2010==
Trevius: Added new quest::resettaskactivity(task, activity) command for resetting a task activity done count to 0.

==06/07/2010==
gaeorn: compile fixes for linux
KLS: Added EVENT_SPELL_EFFECT_TRANSLOCATE_COMPLETE, EVENT_SPELL_EFFECT_BUFF_TIC_CLIENT, EVENT_SPELL_EFFECT_BUFF_TIC_NPC in the similar style of the other spell effect events.

==06/06/2010==
KLS: Added spell quests. Spell quests reside in {EQEMU_DIR}/quests/spells/ and are in the format of {spell_id}.pl. (ex: C:/eqemu/quests/spells/15.pl for greater heal)
KLS: Spell quests allow the overriding of specific spell effect. If a spell quest and sub exists then a spell's effect aside from buff duration and knockback is not used and instead the script is executed.
KLS: Two acceptable subs for spell quests: EVENT_SPELL_EFFECT_CLIENT and EVENT_SPELL_EFFECT_NPC and has a special export of $caster_id.

==06/04/2010=
Trevius: Fixed a potential crash with pick pocket.

==06/02/2010=
KLS: Revert titanium hp boost.
KLS: Random 10 grid type(1) will actually return the 10 closest in 3d space not the 10 closest on the grid list.
KLS: Added new grid type(5): Random 5 with LOS.
Trevius: Added new command #tempname <newname>, to allow temporarily renaming NPCs and Clients. Name resets on repop or zoning.

RECOMMENDED SQL:
utils/sql/svn/1519_zone_primary_key_id.sql

==06/01/2010=
Trevius: Added new Perl Mob quest object TempName(name) to allow changing a name temporarily. Setting no name will return the name to the original name.

==05/31/2010=
Derision: Increased the name field in the internal Ground_Spawn struct from 16 to 20 characters to cater for ground_spawns that are longer than 15 characters.

==05/27/2010==
Trevius: Unreverted the revert from R1502 and added a fix for the tradeskill container combine issue.
Trevius: Fixed a crash when using NukeItem on the Power Source slot for a Titanium client.

PLEASE HELP!

Thanks,

Hunter
EZ Server

[Secrets]

I believe this is resolved in the latest revision (not the one you have.)

Try svn updating and trying again.

I also sent you an email if you need further assistance.

[thepoetwarrior]

There were 13 files updated since I updated a few hours ago, so yup, there has been updates since that hopefully resolve this issue.

Remote control from the Philippines is like using 300 baud modem back in the days lol. Add random brown outs and loss of internet connection randomly, and it sucks.

I'll give the update another chance soon. You'll know its been done when you can zone with over 30k hp =)

Thanks again Secrets!

[Akkadius]

There were 13 files updated since I updated a few hours ago, so yup, there has been updates since that hopefully resolve this issue.

Remote control from the Philippines is like using 300 baud modem back in the days lol. Add random brown outs and loss of internet connection randomly, and it sucks.

I'll give the update another chance soon. You'll know its been done when you can zone with over 30k hp =)

Thanks again Secrets!

Also posted the same thing under the topic "Houston..."

[trevius]

I think this is fixed in Rev1547. It seems like the RNG was producing negative numbers when it shouldn't have for sints and very high numbers for uints, which would cause 100% drop rate.

[thepoetwarrior]

I think this is fixed in Rev1547. It seems like the RNG was producing negative numbers when it shouldn't have for sints and very high numbers for uints, which would cause 100% drop rate.

Thanks for explaining Trevius. I couldn't imagine what the problem was. I was double checking my drop rates, and even looking at the sql files that modified our tables. My last thought was RNG so that explains it. Thanks for fixing quickly!

[thepoetwarrior]

Seems to be working good now. Thanks again for fixing quickly.

[trevius]

Seems to be working good now. Thanks again for fixing quickly.

Glad to hear. I haven't had time to test the latest fix from Gaeorn yet, but it is good to know it works now :)

[gaeorn]

I think this is fixed in Rev1547. It seems like the RNG was producing negative numbers when it shouldn't have for sints and very high numbers for uints, which would cause 100% drop rate.

Actually, the RNG itself only produces uint32 results. The transform to a range of integers (low to high) was where the problem was. But as stated, this has been fixed.