New NWNx4 Persistency retrofit discussion

AcadiusLost · Post by **AcadiusLost** » Wed Jul 04, 2007 7:22 am

Did a first pass at importing these today: guessing you've got a few more files that missed the commit boat.

One seems to be acr_spelltrack_i.nss : the version on the SVN has _playerHasCasterClass(), which you seem to have graduated to a public function, ACR_PlayerHasClass(), in several of the new systems.

I'm also getting an undefined variable or constant, I think on ACR_SERVER_IP in the acr_mod_events_i.nss, not sure where you have that defined in your code.

also:

General question: should we be getting rid of / retrofitting anything that still uses the ACR_GetPCPersistentStorageObject() calls, or do we have those all wrapped to point at the new backend?

ç i p h é r · Post by **ç i p h é r** » Wed Jul 04, 2007 4:06 pm

Rats. I'll commit the spell track code. ACR_SERVER_IP should be defined in acf_settings_i. I'll commit a new placeholder in the repository. Thanks for the sanity checks AL. It's not always easy keeping track of all the modified files when doing a large update. To borrow a phrase from Hugh Grant, sorry for the cock up.

And yes, we do need to completely phase out usage of ACR_GetPCPersistentStorageObject() (though the main include should be left alone - just not used). If you're using that in any spawn code or systems you're working on, it'll need to be replaced with the new persistence functions.

Let me know if you have any questions.

AcadiusLost · Post by **AcadiusLost** » Wed Jul 04, 2007 5:43 pm

Not a problem- I tried plugging a few placeholders in to make sure the rest of it compiled, I think those two were all that was missing from the batch. I still somehow suspect I've missed out committing something somewhere from the last ACR release.

I may give the new code another try tonight if I have time.

ç i p h é r · Post by **ç i p h é r** » Thu Jul 05, 2007 5:33 am

Ok, let me know if you run into any other problems with the latest commit. I think we can consider compiling another release fairly soon. Just to give you a heads up though, the upcoming week is going to be rather hectic at work so I won't be able to spend much time scripting. I'll be in the clear after that.

AcadiusLost · Post by **AcadiusLost** » Mon Jul 09, 2007 7:30 am

Allright, in the early stages of testing, also did the retrofit for acr_xp_i, which also disables the need for validation for now. I'll commit it as soon as I confirm that it actually works as advertised.

When I try to log into my test server, I'm getting booted. The DB creates a record in Players, servers, IP addresses, and CD keys, but not in characters. logs and such show aquire events for the PCs I'm trying, but they're not writing to the "characters" table at all.

I also notice that character ID for all "logs" table entries = 0, so may be something missing from the insert statement for logging.

Thoughts on what might be going wrong?

ç i p h é r · Post by **ç i p h é r** » Mon Jul 09, 2007 7:48 am

Take a peek in your xp_mysql.txt sql log file (wherever you installed NWNx4) and see if any statements are failing. I wasn't getting booted, but then the login script had already created those records for me. My guess is that there may be a syntactical error in one of the sql statements, though I don't recall changing those since I verified them. I did break up the processing into two parts however - OnClientEnter and OnPCLoaded - so perhaps some condition(s) is not being satisfied when OnPCLoaded fires.....

p.s. Are you getting booted on subsequent login attempts with the same character?

AcadiusLost · Post by **AcadiusLost** » Mon Jul 09, 2007 8:49 am

I am getting booted on subsequent attempts. The write to "characters" occurs during the ACR_PCOnClientEnter() function in acr_pps_i.nss, right after the writes to player, CD key, and IP address, so I'm guessing it's that statement that's failing.

Checking that SQL file, there is an error logged that I overlooked:

Code: Select all

! Error (mysql_query): Column count doesn't match value count at row 1.

Not terribly desctriptive, though.

The section of code that tries to write the record is here at line 301:

Code: Select all

        ACR_SQLQuery("INSERT INTO characters (ServerID, PlayerID, Name, Level, Race, Subrace, Deity, Gender, HP, XP, GP, Ethics, Morals, Class1, Class2, Class3, Level1, Level2, Level3, STR, CON, DEX, `INT`, WIS, CHA, Damage, Deaths, IsOnline, IsDeleted, IsPlayable) VALUES (" +
	 	IntToString(ACR_SERVER_ID) + "," +
           	sPID + ",'" + 
           	sCharacterName + "'," + 
           	IntToString(GetHitDice(oPC)) + "," +
           	IntToString(GetRacialType(oPC)) + "," +
           	IntToString(GetSubRace(oPC)) + ",'" +
           	GetDeity(oPC) + "'," +
           	IntToString(GetGender(oPC)) + "," +
		IntToString(GetCurrentHitPoints(oPC)) + "," +
           	IntToString(GetXP(oPC)) + "," +
           	IntToString(GetGold(oPC)) + "," +
           	IntToString(GetGoodEvilValue(oPC)) + "," +
           	IntToString(GetLawChaosValue(oPC)) + "," +
           	IntToString(GetClassByPosition(1, oPC)) + "," +
           	IntToString(GetClassByPosition(2, oPC)) + "," +
           	IntToString(GetClassByPosition(3, oPC)) + ",1,0,0," +
           	IntToString(GetAbilityScore(oPC, ABILITY_STRENGTH)) + "," +
           	IntToString(GetAbilityScore(oPC, ABILITY_CONSTITUTION)) + "," +
           	IntToString(GetAbilityScore(oPC, ABILITY_DEXTERITY)) + "," +
           	IntToString(GetAbilityScore(oPC, ABILITY_INTELLIGENCE)) + "," +
           	IntToString(GetAbilityScore(oPC, ABILITY_WISDOM)) + "," +
           	IntToString(GetAbilityScore(oPC, ABILITY_CHARISMA)) + ",'" + "',0,0,1,0,1)");

Could it be that it's trying to pass a string for ServerID instead of an integer?

ç i p h é r · Post by **ç i p h é r** » Mon Jul 09, 2007 3:18 pm

AcadiusLost wrote:Checking that SQL file, there is an error logged that I overlooked:
Code: Select all
! Error (mysql_query): Column count doesn't match value count at row 1.

What's the preceding SQL statement or statements?

Could it be that it's trying to pass a string for ServerID instead of an integer?

Well, if ACR_SERVER_ID is a number and the column is an integer (which it is), then MySQL treats it as an integer. To generate an error, I think you'd have to try to pass a string to an integer column w/o quotes.

Easiest way to find where it's failing might be to just comment out the boot statements one at a time.

AcadiusLost · Post by **AcadiusLost** » Mon Jul 09, 2007 4:15 pm

Unfortunately, the xp_mysql.txt doesn't have a preceeding SQL statement: here is the file in it's entirety.

Code: Select all

NWNX MySQL Plugin V.0.0.1
(c) 2006 by Ingmar Stieger (Papillon)
visit us at http://www.nwnx.org
(built using mysql-5.0.27 source)

* Connecting to server x3mkungen.mine.nu
* Plugin initialized.
* Registering under function class SQL
! Error (mysql_query): Column count doesn't match value count at row 1.
* Plugin unloaded.

I can only guess that it's the insertion line into "characters", as the preceding one seems to have written to the DB successfully. I'll try commenting out the boot command associated with it and see if that lets us get further along.

ç i p h é r · Post by **ç i p h é r** » Mon Jul 09, 2007 6:51 pm

Ok. Check your nwnx.ini file (or maybe it's mysql.ini) and make sure the log option is set to 2. That will log everything. At that setting, you should see all the SQL statements executed by the ACR in your sql log file along with their return values and any associated warnings or errors. This output will obviously help debug our sql but it'll also be a good way to identify potential optimizations.

AcadiusLost · Post by **AcadiusLost** » Mon Jul 09, 2007 7:02 pm

I know the log option is set to 2 in at least one of those files, but I'll check them all when I'm home tonight.

AcadiusLost · Post by **AcadiusLost** » Tue Jul 10, 2007 5:07 pm

Logging was set to 2, but still no feedback for whatever reason.

Commented out that line, but now it appears the SQL server is offline, so we can't test in any case.

ç i p h é r · Post by **ç i p h é r** » Tue Jul 10, 2007 5:15 pm

I can take a look at it at the end of the week. Were you able to narrow down which condition was responsible for the boot?

I need to talk with Zicada again and nail down our hosting plans.

AcadiusLost · Post by **AcadiusLost** » Wed Jul 11, 2007 9:52 pm

SQL server's address is pinging again, so may be able to resume testing tonight. We'll need to make sure we have a robust server (and maybe a backup solution in place) before ALFA goes live for NWN2, down the line. At current, SQL DB downtime means everyone gets booted from all NWN2-ALFA servers.

AcadiusLost · Post by **AcadiusLost** » Thu Jul 12, 2007 6:35 pm

Did some more testing and tinkering with this late last night.

In no case is it writing records to the "characters" table. The other tables seem to be updating fine. I was able to join the module with a PC by commenting out two of the bootPC lines: the one checking the insertion into characters onPCClientEnter, and the one checking the query from characters OnPCClientAreaLoaded. Couldn't test much in terms of other systems, as I expect most of them need to reference the characters table. Had a careful look over the INSERT syntax, couldn't see anything wrong with it. The "logs" table records the errors in loading the characters, as well- so there has to be something going wrong with that insertion. The NWNx4 log file still just shows that same error from before (albeit in repeat for each time I tried).

The one thing I didn't try, was to manually enter one of my PCs into the characters table with the SQL browser, and then see if it worked for other systems (or updated properly for PC death and such).

Cipher, can you try rolling a PC under a new GSID (or even your normal one) to confirm that it's still working on your end?