Disc & Total healing Issue

[Tmr]

Hi,
It seems that many spells are incorrectly calculated right now (It took a few days on french forums to identify the source, which happens to be Wol)

issue 1 : Atonement
illustration with http://www.worldoflogs.com/reports/ju03 ... 756&e=1159
Attonement has two different lines, on the "healing by spell --> healing taken" section
--> healing is 11796 on this player
--> healing is 8827 on this player

a log query gives
[10:51:04.015] Tmr Atonement Holivertwist-Cho'gall +*0* (O: 27200)
[10:53:20.093] Tmr Atonement Holivertwist-Cho'gall +8827 (O: 5994)
[10:53:45.457] Tmr Atonement Holivertwist-Cho'gall +*11796* (O: 18036)

It means that Atonement healing is not counted as one spell, but two. total for those two spells :
http://www.worldoflogs.com/reports/ju03 ... 756&e=1159
http://www.worldoflogs.com/reports/ju03 ... 756&e=1159

But only one of these two spells are counted on the "healing done" page
http://www.worldoflogs.com/reports/ju03 ... 756&e=1159

Rq: Recount AND skada gives the correct amount of absorb, which happens two be the exact sum of the two spells.


Issue 2 : Divine aegis
(same log, same fight analysed.)

http://www.worldoflogs.com/reports/ju03 ... 756&e=1159
A holy priest has some healing done via Divine Aegis. It implies that the holy priest healing done is over-evaluated, and the disc's under-evaluated


Issue 3 : Spirit Shell
(same log, same fight analysed.)

http://www.worldoflogs.com/reports/ju03 ... 756&e=1159
Spirit shell is not counted on anybody exept the tanks. A query on one DPS gives :

[10:52:01.213] Kyary-Kael'thas gains Spirit Shell from Tmr (Remaining: 31811)
[10:52:16.209] Tmr's Spirit Shell fades from Kyary-Kael'thas (Remaining: 27517)
[10:55:43.475] Kyary-Kael'thas gains Spirit Shell from Tmr (Remaining: 30368)
[10:55:58.448] Tmr's Spirit Shell fades from Kyary-Kael'thas (Remaining: 25388)

It looks as if Spirit Shell is properly evaluated by WoL only if the full amount of absorbs is consumed (ie on the tanks), not if it fades by timer.

[Maihem]

I'll look into it.

[Maihem]

Oops. ArcaneMagus just pointed out that I'd forgotten to post a follow-up to this topic. Since we haven't really changed anything in the heal/absorb code, I'll assume this is still an issue.

[Issue #1]

Looking at Holivertwist-Cho'gall: Healing Taken by Spell, there are two Atonement entries, #94472 for 11796 and #81751 for 8827. This is reflected in the two Spell Details pages you linked in their respective Healing Taken by Holivertwist-Cho'gall entries. I'm not entirely how the Tmr: Healing Done by Spell page is wrong here as it does actually show two Atonement spells in the Healing Done by Spell table. You can verify that these spells are both taken into account by looking at the Analyze Healing Done page, selecting 'by spell', source=Tmr, target=Holivertwist-Cho'gall. This shows two Atonement entries with the aforementioned values 11796 and 8827 which seems correct to me.

[Issue #2]

Holivertwist is shown to have healed using Divine Aegis because that's what the combat log says. His Divine Aegis was applied twice with a value of 6689 and faded twice, which means he has 2x 6689 shield healing, matching the numbers on the page you linked. See details here:

http://www.worldoflogs.com/reports/ju03 ... %27gall%22

[Issue #3]

Looks like we aren't counting absorbs correctly in all cases. Specifically, when a shield fades with a remaining value greater than zero, WOL would count this as full overhealing. Instead, we should be counting the difference as effective healing since that has been "consumed" by absorption in the mean time.

I have committed a patch to remedy this issue and it seems to be working on the development environment. We'll push out this patch to the live servers soon, hopefully today.


My apologies for taking so long to investigate this issue properly -- kinda forgot about it

[Badmissery]

I would like to add something to this topic.
Is it possible that WOL can't count the absorb of stackable buffs?

Best example is a disc priest with "Spirit Shell". I activate it and cast a big heal on the tank, he will get a shield also called "Spirit Shell" for maybe 60k. Now he gets damage for 40k so he still get a shield for 20k. Now I cast my next big heal and add 60k to the existing 20k so he will get a shield for 80k absorb.
Is it possible for WOL to track this shields accurate?

Log Example:
http://www.worldoflogs.com/reports/rt-nad5g0bfc2r8vwqx/details/2/?s=2506&e=2868
Healing by Spell shows me that my Spirit Shell absorbed 460k damage

But the log said
http://www.worldoflogs.com/reports/rt-nad5g0bfc2r8vwqx/xe/?s=2506&e=2868&x=spell%3D%27Spirit+Shell%27+and+sourceName%3D%22Badmissery%22
The shield was capped and I know the tank gets damage in this time, so it had to be much more that 460k.

Can you tell us how you track absorb at all? And for which player this absorb will count as a heal?
And exist this problem for all stackable absorbs?

Thanks for your time.

[Maihem]

Yeah, we noticed this yesterday as well. It looks like Blizzard has implemented a change in shield handling since Cataclysm.

When we last changed shield tracking, WoW would fire a SPELL_AURA_REFRESH event each time the shield value changed, essentially using it as a "shield update" event. You can see an example of that at the link below. This means that each time the shield value changed due to an absorb, we would count the difference toward healing.

Shield update events example: http://worldoflogs.com/reports/rt-1ez5o ... d+%3E+0%29

Now, however, these intermediate events no longer seem to fire. This means that WOL's shield tracking no longer works correctly in the case you outlined. The absorb inbetween will no longer be counted in the new situation. Before, it would have been like this:

GAIN (value=60k)
<hit for 40k>
REFRESH (value=20k) --> 40k healing
<heal that adds 60k to shield>
REFRESH (value=80k) --> default 20k overhealing unless we know it's a stacking shield, in which case it's 0 (over)healing, just an update
<...>
FADE (value=X) --> 80k - X healing, X overhealing

We'll need to think of how to handle this new situation, but I'm afraid that tracking stacking shields correctly has become impossible with this change =\