Some new feedback on the pet threat problems being reported by a variety of hunters. Since Hortus's testing in the previous thread ruled out some of the possible problems, I decided to focus on what I considered to be the most likely culprit: Growl being starved of focus.

Now, as far as I know, Growl was special cased prior to 2.4 so that if the pet had more than 15 focus, Growl was on autocast and not on cooldown, the pet would cast Growl, and only then check to see if there was enough focus available to autocast something else. My theory is that this special casing has been lost, and other abilities (like Bite and Gore) are getting cast without leaving enough focus for Growl. I vaguely remember this being explicitly added somewhere along the line (I think Growl may have had priority pre-TBC, got broken in 2.0, then fixed in one of the many 2.0.x bug fix builds).

As far as I know, the pet also used to happily cast Growl even if the mob was immune to it - Hortus's posts in the previous thread suggest that is no longer the case, further suggesting there were changes to this area of the code in 2.4, and hence the chance of breaking things.

I initially came up with this theory because of the various reports coming from hunters that farm with boars that melee or other special attacks made by the pet were sometimes consuming the Charge buff before the first Growl had the chance to benefit from it (the charge buff results in 3000+ additional threat from Growl).

So I did some testing on Darkspine Myrmidons, and I found that the delay between Growls was frequently a lot more than the 5 second cooldown. My ravager had Bite, Gore and Growl all on autocast for my testing and there would often be cases where the pet would go for 8-10 seconds just using Gore before Growl finally triggered again.

Once I actually started firing and Go for the Throat kicked in, then the pet had no trouble firing Growl off every time the cooldown was up (granted, very limited testing of that since the naga were generally almost dead by the time I opened up).

One very nice thing about this theory is that it explains the reports of pets taking "2 or 3 growls" to get aggro back after losing it to the hunter: it isn't actually 2 or 3 Growls, its just taking that long for the pet to get the focus together for 1 Growl.

A few things to consider that may govern whether or not different players and test scenarios are affected by this (assuming I'm correct about the problem):
1. Does the hunter use a pet with a focus dump on autocast while soloing? If no, they probably won't notice any change between 2.3 and 2.4 (since their pet is going to be running around with a near full focus bar most of the time regardless)
2. Does the hunter have the Go for the Throat talent, and is the hunter firing? If yes, there probably won't be any change between 2.3 and 2.4. This goes directly to the heart of Hortus's testing in the other thread: since the description of the test indicated the testing hunter was firing, did that hunter also have GffT trained?
3. Does the hunter have Bestial Discipline trained? If no, I think focus may come in small enough ticks that it is likely that Growl will get a shot at the focus pool when it is above 15 but below the 25 needed for Claw or Gore. With BD trained, it is quite possible to skip right over that range and trigger Gore instead of Growl.


Alternate Speculation:
Prior to 2.4, I can't recall my pet's focus bar getting to zero anywhere near as fast as it was during my testing this evening.

Given that 2.4 added the change to allow haste to affect spell casting GCD and pet abilities are considered spells cast by the pet, is it possible that Frenzy, Serpent's Swiftness and Cobra Reflexes are lowering the pet's ability GCD, such that spammable abilities like Gore and Claw are now consuming way more focus-per-second than they used to? I was only looking at the time delays between Growl in my combat log - maybe it would be worthwhile looking at the time delay between successive Gores with everything else switched off.

Hortus, did the hunter in your testing have Frenzy and/or Serpent's Swiftness? Did the pet have Cobra Reflexes?

---

Q u o t e:
Some new feedback on the pet threat problems being reported by a variety of hunters. Since Hortus's testing in the previous thread ruled out some of the possible problems, I decided to focus on what I considered to be the most likely culprit: Growl being starved of focus.

Now, as far as I know, Growl was special cased prior to 2.4 so that if the pet had more than 15 focus, Growl was on autocast and not on cooldown, the pet would cast Growl, and only then check to see if there was enough focus available to autocast something else. My theory is that this special casing has been lost, and other abilities (like Bite and Gore) are getting cast without leaving enough focus for Growl. I vaguely remember this being explicitly added somewhere along the line (I think Growl may have had priority pre-TBC, got broken in 2.0, then fixed in one of the many 2.0.x bug fix builds).

As far as I know, the pet also used to happily cast Growl even if the mob was immune to it - Hortus's posts in the previous thread suggest that is no longer the case, further suggesting there were changes to this area of the code in 2.4, and hence the chance of breaking things.

I initially came up with this theory because of the various reports coming from hunters that farm with boars that melee or other special attacks made by the pet were sometimes consuming the Charge buff before the first Growl had the chance to benefit from it (the charge buff results in 3000+ additional threat from Growl).

So I did some testing on Darkspine Myrmidons, and I found that the delay between Growls was frequently a lot more than the 5 second cooldown. My ravager had Bite, Gore and Growl all on autocast for my testing and there would often be cases where the pet would go for 8-10 seconds just using Gore before Growl finally triggered again.

Once I actually started firing and Go for the Throat kicked in, then the pet had no trouble firing Growl off every time the cooldown was up (granted, very limited testing of that since the naga were generally almost dead by the time I opened up).

One very nice thing about this theory is that it explains the reports of pets taking "2 or 3 growls" to get aggro back after losing it to the hunter: it isn't actually 2 or 3 Growls, its just taking that long for the pet to get the focus together for 1 Growl.

A few things to consider that may govern whether or not different players and test scenarios are affected by this (assuming I'm correct about the problem):
1. Does the hunter use a pet with a focus dump on autocast while soloing? If no, they probably won't notice any change between 2.3 and 2.4 (since their pet is going to be running around with a near full focus bar most of the time regardless)
2. Does the hunter have the Go for the Throat talent, and is the hunter firing? If yes, there probably won't be any change between 2.3 and 2.4. This goes directly to the heart of Hortus's testing in the other thread: since the description of the test indicated the testing hunter was firing, did that hunter also have GffT trained?
3. Does the hunter have Bestial Discipline trained? If no, I think focus may come in small enough ticks that it is likely that Growl will get a shot at the focus pool when it is above 15 but below the 25 needed for Claw or Gore. With BD trained, it is quite possible to skip right over that range and trigger Gore instead of Growl.


Alternate Speculation:
Prior to 2.4, I can't recall my pet's focus bar getting to zero anywhere near as fast as it was during my testing this evening.

Given that 2.4 added the change to allow haste to affect spell casting GCD and pet abilities are considered spells cast by the pet, is it possible that Frenzy, Serpent's Swiftness and Cobra Reflexes are lowering the pet's ability GCD, such that spammable abilities like Gore and Claw are now consuming way more focus-per-second than they used to? I was only looking at the time delays between Growl in my combat log - maybe it would be worthwhile looking at the time delay between successive Gores with everything else switched off.

Hortus, did the hunter in your testing have Frenzy and/or Serpent's Swiftness? Did the pet have Cobra Reflexes?

---

Interesting...this sorta matches what I have noticed...

Which proves more that Hortus test was not exactly a "true" test. The conditions under which he did the tests would not show any Focus or lack of focus influences.

I reproduced as best I could, the so called test he did...and sure enough I never drew aggro...of course I did not have threat numbers like he did...but who cares..the fact is i never drew aggro.

But if I go out and play like Hunters actually do...then there are aggro issues...and much different pet aggro behavior than pre 2.4

Alot of what you say could very well be happening....they changed alot of stuff in 2.4...given their record for breaking more than they fix with every change I would not be surprised if the mecha

---

Q u o t e:
Which proves more that Hortus test was not exactly a "true" test. The conditions under which he did the tests would not show any Focus or lack of focus influences.

---

It was a true test of what he was trying to test at the time - whether there had been any changes to the underlying threat generated by Growl or Auto Shot. His testing ruled that out quite nicely, making it clear that any problems had to be somewhere else.

Something that actually occurred to me in relation to the possibility of pet GCD being affected by haste talents and skills: if this was actually happening, it could easily break the Growl prioritisation as the spammable ability could be coming off CD slightly before Growl and stealing the focus Growl would have otherwise used. So when the Growl CD finishes, the pet is still waiting around for the next focus tick before it can cast it.

If that proves to be the case, then not having BD would actually make the pet's threat generation much worse, as it would sometimes require an extra tick to get enough focus for Growl.

More things to try out this evening :)

Nice post Nimizar - hope to hear some more about this soon.

Thanks for the great post Nimizar, hopefully you can track down what is happening.

I thought that I would add to your post by presenting the other problems with pets:

- I'm pulling agro far too easily now in 2.4 and far easier than in 2.3.3
- My pet seems to take far more damage than it use to under 2.3.3 causing me to use more mana hastening my mana shortage
- My pets always seem to be hungry now. I know I am feeding them far more often but they still get manage to get hungry far too quickly.

The first issue I'm unsure what has changed but something definitely has. My pet now has a very hard time holding agro off me where before it was easy to balance the threat (and play agro ping-pong). I don't know if it is just me doing more threat (unlikely) or the pet doing less (more likely). It's a pitty we don't have access to a 2.3.3 server so we can give Hortus some evidence of the changes. I'm just hoping that Nimizar or someone can track it down.

I think the third issue may have something to do with the increase in the level 70->71 xp. They recently increased the xp required to reach 71 from 814,700 XP to 1,256,000 XP. Mania mentions (http://www.maniasarcania.com/2008/03/31/patch-24-level-70-xp-increase/) that your pet's loyalty is based on your XP could it be the increase to the level 71 XP that is causing our pets to lose loyalty faster?

While I realise for Blizzard that these issues may be hard to track down, something is definitely going on with hunters. I'm really hoping that you can fix them soon. At the moment I've gone back to the warrior since my hunter pets are all broken :(

Edit: grammar

Spot on observation. I read somewhere that Growl had its priority changed, this reduces the Charge aggro since it no longer boosts the Growl, but rather the autoattack. As a workaround, is there a way to turn off autoattack on the charge? Via either addon or macro?

---

Q u o t e:
Spot on observation. I read somewhere that Growl had its priority changed, this reduces the Charge aggro since it no longer boosts the Growl, but rather the autoattack. As a workaround, is there a way to turn off autoattack on the charge? Via either addon or macro?

---

I think a change in Growl priority is still only a theory at this point - any change that led to Growl being starved of focus would lead to similar symptoms when chain pulling mobs.

I haven't really had a chance to get back and do any more systematic testing, but running dailies for a bit with Gore switched off seemed to give a consistent 5-6 seconds between Growl entries in the combat log - none of the 8-10 second outliers I saw when I had Gore switched on.

I am going to try using only growl and no bite/claw/gore on my boar, cat and wolf. If this stopgap helps, it will at least be something.

P.S.:

I didn't have issues with focus before 2.4 at all, I have bestial disc and GffT trained.

OK, had a chance to make a few more observations this evening, using the Wretched Fiends/Devourers on sunwell as my guinea pigs. All of the observations below are based on the period before I started firing, so the pet was relying only its native focus regen (including the effect of Bestial Discipline). No additional focus was available from Go for the Throat.

Step 1 was to set up a filter in the new combat log that showed only spell casts by my pet, and included timestamps in each of the messages. Even if the events are timestamped client side rather than on the server, the variation in the time it takes the spell cast messages to get from the server to my machine should be well below the 1 second granularity of the log.

Step 2 was some testing with only Bite and Growl on autocast - this showed Growl going off every 5-6 seconds pretty reliably, but when the Ravager first attacked the mob, Bite was sometimes listed first. This suggests we have a prioritisation problem when it comes to focus expenditure, rather than it being anything to do with Gore now consuming more focus-per-second.

Step 3 was some testing with only Gore on autocast - this showed quite a few sequences with Gore being cast every second until the pet ran out of focus. After that, Gore would be cast whenever focus was available.

Step 4 was to go back to having all 4 abilities (Dash, Growl, Gore, Bite) on autocast. With this setup, it was quite common for either Bite or Gore to be listed before the first Growl. The absolute kicker though was the one occasion where the sequence went Dash, Bite, Gore, Gore - that's right, ZERO Growl casts before going into focus starvation mode with Gore and Growl competing for focus.

I tried to have a bit of a look through WWS logs from before and after the patch to see if I could spot any systematic differences but there's too much noise in the values from external factors like changes in my gear and whether or not I had a feral druid or enhancement shaman in my group (more crit -> more focus for the pet -> greater fraction of damage from Gore).

From my observation of the combat log, it looks like the prioritisation system for pet ability autocasts has been completely hosed:
1. Higher focus cost abilities are not being given priority, as I saw plenty of cases where Gore appeared in the Blizzard combat log before the first Bite)
2. Growl is not being given priority over everything else, as not only did I see plenty of cases where Bite or Gore appeared in the Blizzard combat log before the first Growl, I also found that the time between Growls was often up around the 8-9 second mark when it was competing with Gore for focus, as opposed to the 5-6 seconds seen when Gore was taken off autocast.

"But wait," you say. "Hortus tested this and didn't find a problem! How do you propose to explain that?"

---

Q u o t e:
I have been able to verify by side by side comparison that Growl auto-casts consistently when turned on and that it is doing the exact same amount of threat in 2.4 as it did in 2.3.3.

I also did side by side testing of two identically geared and talented hunters (using a copy of a retail character that reported this problem) with identical fully skilled pets and compared threat.

I sent the pets to attack an unkillable target.
After the opening growl I turned on Auto-shot, and hit Arcane Shot.
I then chain cast arcane shot 9 more times as soon as it was off cooldown.
I then compared the threat of the pet and the hunter. Both were within 10% of each other.

I repeated this on 2.3.3 and 2.4 and the results were nearly identical.

I also repeated this test using Improved Hunters Mark (as I notice all the hunters reporting it have that talent) and the threat for both the pet and the hunter were higher but still within acceptable margins and aggro was not lost by the pet by the 10th arcane shot on 2.3.3 or 2.4.

There may be a bug with hunter pet aggro, but it is not caused by Growl. If anyone has more information or observations about threat they can post here and I'll try to take a look at them.

---

What Hortus doesn't tell us here is whether or not he had anything other than Growl on autocast. Without Gore on autocast, I see results similar to his (although I would probably have to be using my boar or wait until the second Growl to get through that many arcane shots without pulling aggro, even before the patch).

Nice work!!

We really need someone besides Hortus testing this. The shot rotation was laughable at best.

WTB someone at Blizz that actually knows something about the hunter class...

THIS is what type of testing should've been done from Blizzard's side. Seems to me that with their "internal programs" they could track down the exact process that's breaking.

I have been running my cat with only growl on autocast for the day. I did notice a big improvement, but something is still not quite right. I can still pull aggro off the pet with only white damage at times. I have also been on the lookout for resists and have not noticed any. With claw on, I had the mob in my face almost every time, so I think you found a big part of the issue. Well done!

Might as well stop the testing folks, Hortus confirmed on another thread. Broken as Intended.

I have confirmed with the developers that Boar's growl ability getting such a large bonus to attack power based on charge is a bug. Growl should only benefit from the hunter's attack power and get no benefit from buffs to the pet's attack power.

Expect this to be changed in a future patch. If you disagree with this change please post about it in the suggestions or class forums.

http://forums.worldofwarcraft.com/board.html?forumId=11122&sid=1
[ Post edited by Hortus ]

---

Q u o t e:
Might as well stop the testing folks, Hortus confirmed on another thread. Broken as Intended.

---

Maybe the charge change is intended, but from what I'm seeing, casting priorities were also changed
in a way that is really hurting hunter aggro. That's probably a bug, and people should be asking to
have it fixed, in my oppinion.

I have a hypothesis on this problem.

When 2.4 came out, it reset or perhaps re-arranged the order of the pet [auto]cast bar. It definitely did this for me. I always used to have Growl() as the left-most item. When 2.4 came out, suddenly it was the right-most item on the pet bar. I used the ordering as my identifier for what was auto-casting, so I clicked off Claw() and clicked on Growl() by accident in a raid on Tuesday. That's why I know the order was changed.

There must be some order of expended focus. Growl takes 15 focus. If focus is *initially* expended in left-to-right pet bar order then Growl will occur less frequently. After initial expenditure of focus, it will be based on cooldowns and availability of focus.

To try the hypothesis, drag and drop the auto-cast-bar icons to re-arrange them in your original or the possible bug order. Otherwise, the ordering may have been changed under-the-covers and not in any way accessible to us to test.

Hortus' post was only about the boar's charge/growl threat.

Every pet out there is having a growl/threat problem, not just the boar.


Good post, keep getting data!!

---

Q u o t e:
Dear hunters, we have identified a bug where one of your pet families was actually a semi-effective tank, able to stop a mob in its tracks. This was unintended, and thus we're going to break all threat generation for all pets just to nerf this nice bonus for your class that wasn't actually hurting anybody.

---

(Bitter? Moi? Never...)

C'mon Blizz, could we get at least a little communication about what the hell you're trying to do with our class? Apparently we were too good at soloing, so you felt it necessary to nerf our non-casting mana regen and our pet's threat into the ground.

I'm done with my testing on this by the way: when I had all of Gore, Bite and Growl active, the order the 3 would activate when my ravager first attacked a mob was pretty random, suggesting we can't control this by putting things on the pet bar in a different order.

Since I don't have any other ideas on things to test, it's all in Blizzard's court now. Hey, maybe we can get the developer back that implemented the pet AI changes so they would duck cleaves in boss fights? At least that change was made with the clear intention of trying to make our pets more effective (even if it took a while to work the kinks out), but Blizz haven't given any indication of why they touched pet autocast mechanics in the latest patch at all.

/end derail (maybe someone else will think of additional things to test)

Bump for the hope that developers/debuggers actualy LOOK at this and try to figure out why someone NOT on thier payroll is doing their job for them.

Great testing btw.


-Cut

You want the truth! You can't handle the truth!