Just because people prefer pain to death doesn't mean the pain is not a problem. You can easily present me with a choice of two options, and i would confidentially choose one because it is much better, but would also strongly prefer not having to choose either of them.
There is no full guarantee that they wouldn't die if they were not fighting wraiths. Or that a fight would happen at all.  And "but what would happen otherwise, maybe if you did nothing at all" is a complicated kind of approach/calculation that, while i admit might be useful for some very smart entities, is not practical in my case.

There are no "eager" volunteers. Just "those willing to endure, to a smaller or larger degree". Having enough is a complicated question, it's not a binary "agree/not agree". If there are no other options, i am willing to provide all the energy, no matter how much is needed.  So in one sense it can be said that there is no problem at all, because it means no one else needs to make that sacrifice, all the suffering can be inflicted on only me. But i would prefer not to do it alone, for all my altruism (i think this is a deontologically acceptable perspective, even not counting "damaging my mental state that way would on net reduce the amount of utility i can produce"). There probably are economic, psychological and mathematical terms and theories relevant here, about the relative preference for things, though i don't remember them precisely.
There are also the essentially impossible questions of quantifying suffering. Can some amount of pain can be said to be precisely twice larger than another? Is it better (all else equal) for one person to experience two units of pain, or for two people to experience one unit each? Maybe spreading the pain around will be worth it on net despite the additional possibility of psychological trauma (which is not a punishment and not intended to change anyone's behavior). Or maybe not.

Creating more wraiths is essentially impossible (it uses a resource that i know no way of acquiring more of). We are experimenting with creating simpler and weaker forms of undead, with the possibility of converting everyone at some point. But all types have their own disadvantages, and undead are fundamentally not able to repair their bodies when damaged (except for the highest undead that can do that by draining life from others, like vampires and wraiths).  We are working on workarounds for that too, but there is no way to tell yet if that is possible at all.
(If the concern about "the need to kill moral agents to survive" is about meat, i will note that life magic can theoretically be used to affect nutrients bodies need, and/or modify plants to have different nutrients. That is not my specialty, and not currently seen as a priority by anyone.)

The discourse has been rational so far, it seem worth continuing.

Perhaps we should engage in some more "semantics", it seems I have miscommunicated somewhere...

First of all, I never advocated for "slavery".  I used the term "compel" colloquially, but in fact what you would be doing is offering them an additional option to pay back the damage they caused.  The fact that the alternative payment options involve immense amounts of pain is not a game theoretic threat on your part, it is your best alternative to a negotiated agreement.  Many moralists would wring their hands about some abstract sense of slavery or coercion, but I fail to see how providing people additional options can in any way be "enslavement" or "coercive" in the truest, most logical sense of the words.

As to my usage of the term moralist... I've picked up that you are still relatively young?  As you experience life you will encounter many people that try to appeal to arbitrary rules and felt senses that in fact correspond poorly to reality.  I have encountered many such people.  Occasionally I can make some slight progress in persuasion when they agree to civil discourse, but often it breaks down into name calling and insults from them.  They claim to be motivated by "morality", and I in fact agree that one's core utility function is not up for compromise, but in fact the moralists take many actions that are suboptimal for fulfilling their alleged values and fail to take many options that would greatly support their alleged values.  If anything, in terms of efficiently providing value to society, I greatly surpass them, so I would claim that from a logical, utilitarian perspective of morality I am the more moral person.  But I digress.  If you are truly motivated by morality, following my advice will be quite useful for optimizing for your moral values!  I use the term "moralist" to disparage those that follow an arbitrary, false, self-righteous, contradictory "sense" of moral rules (instead of a logically, consistently self-articulated value function).

As to the other commenter you responded to... if a moralist is so far gone as to consider an animal as a moral patient there is likely no appeasing them. On a side note, if, for some bizarre reason, animal suffering actually factors into your value function, I would recommend a campaign of mass extermination.  Proper large scale necromantic rituals can exterminate animal life painlessly and swiftly, thereby minimizing animal suffering.  I do not have this value myself, but once offered to design such a ritual as payment to someone I thought was rational enough to at least consider the offer in good faith (they promptly ended our previously fruitful discussion with pointless ad hominins).

Well, let’s imagine a situation:
You need labor, but don’t have enough of it, and you don’t actually want to pay for it, or can’t. But you have a powerful and skilled army. And so you attack someone, and enslave them, for your own economic benefit.
Let’s say you can't do that, because there is a very important principle, that the army only exists for defense, and never attacks first, and can only compel to work people who have hurt you during war. But you still have a powerful army. And you still have an incentive to get the labor for free. So you will have an incentive to provoke sides (which are weak enough to easily defeat) into technically attacking you first, and them compel them to work, as compensation for harms they technically caused you (your incentive to interpret everything as harm also becomes large) legally and without officially breaking the principles.
Is it slavery? Maybe not. I called it strategies of the "enslave defeated enemies" type. It doesn’t matter what I call it, the situation will remain the same (which what "getting into semantics" usually means), an outcome which is opposed to "the general efficiency of the world", or "the project of civilization", or whatever people want to call it. 
And providing people options in a situation can be coercive if you are the one who put them in a situation in the first place (which not necessarily be the case. But it’s a possibility worth remembering).

Believe me, I have enough experience with people incoherently believing some completely normal thing to be inherently and overwhelmingly immoral for reasons they can’t justify. And I can distinguish those cases from coherent moral positions relying on valid arguments.
That's why i don’t call them "moralists", I call them "idiots" or "fanatics", or "naïve deontologists" (as opposed to rational deontologists) if I am trying to be somewhat polite. Though, again, that’s semantics, the name doesn’t matter if we agree about the actual phenomenon.

The idea of exterminating life to prevent suffering is internally coherent, but flawed, because it assumes a very simple meaning of the term "suffering", which is not the meaning most relevant to moral arguments (though whether animals are capable of the relevant type of suffering is questionable. I don’t have a strong opinion on it, myself, focusing on higher priorities).

[info] does you society have the concept of anesthesia? If not can you find some of these?
 

Anesthesia affects the body, and pain coming from the body. It doesn't, as was mentioned earlier, help with the feeling of vital energy being drained out of your soul.

You have people willing to sacrifice themselves to support your civilization? You do, because you have an army.

Allow them that sacrifice. Take volunteers, or order your soldiers, for a set of people, perhaps one per wraith, who will shield all the rest. 

You can heal people, and so those you choose can sustain the wraiths as long as they live. If their minds cannot bear it, break or erase their minds. They no longer serve a need. 

Note to self: double check that the Communicator doesn't allow any magical or mental contact to the entities on the other side.

Would it be possible to drain animals that would be being killed anyways? I'd also want to see if you could make it work with brain dead people - some maladies destroy the mind but leave the automatic functions of the body still running, and if you have enough healing capacity, it seems like it could be a workable solution without causing sapient suffering persay. 

This is a very good question!
It doesn't work on animals. Animals are not sapient, and that means their souls are not..."structured" enough, or "saturated"? To be effectively drained.
But, does that apply to creatures that were sapient, had those kind of souls, but then lost their capability for thought? I genuinely have no idea!
The energy drained is vital energy, not mental, so technically speaking, it shouldn't matter. Though maybe lack of feeling and not just thought would be serious enough to make the soul no longer fully function...But there is not really any knowledge on that in practice. That would require research.
Still not the most ethical research. Not like we have a lot of brain-dead patients. But killing those we were going to kill anyway is not that much worse, in the large scale of things.

There’s a complementary question and an experiment that question naturally suggests!  Do creatures that will gain the capacity for sapient and thought have souls?

The obvious follow-up question and experiment: Does the ritual work on babies?  Now, I don’t know if you have any way to check without trying the ritual on a baby… Babies present a complicated ethical challenge, as they are (seemingly) not yet economic agents able to consent, but they will become economic agents in the future.  You could try to predict how much payment they would want it/when they consent.  (Have you heard of prediction markets?  They are an excellent tool for making fair future estimates!).  Alternatively, if your society has the moralist constraint of considering children under the protection of their parents, you could try compensating their mothers instead!  

For a clever third option: advanced telepathic magic may allow you to acquire consent even if babies aren’t yet verbally communicative.  (And for added redundancy you can compensate the parent(s) as well.)   This is my preferred solution to rituals that work best with babies.  (I personally have found a few surprising results where rituals that you think would work best with a fully developed soul actually work best with souls with the most potential, I.e. baby souls!  As an added bonus, although babies often have tricky abstract requests, on net providing these requests and a payment to parents is often far cheaper than compensation for equivalent ritual roles to adult employees!) 

Oooh, more opportunities to talk about thaumatology!

"What a soul even is" is a complicated question. Some even classify physical bodies as part of the soul, "the first layer".
I am part of the approach saying that a soul is the astral structure associated with a living body, so bodies don't count, and neither do inanimate objects, which merely have "astral shadows", but not real souls.
Plants, however, do have souls, just very simple souls, lacking a lot of the "higher" components sapient creatures have.
[Several paragraphs of confusing details, clarifications and disclaimers omitted]
That has nothing to do with morality, though, the lack or presence of a soul is not what i consider to be the reason of being a moral patient. Neither is the ability of becoming an economic agent, really, that affects the actions you can take, not the actions you should take or those that would have been beneficial if you could do them.
What's important is the capacity to experience suffering and happiness, because meta-morality is the goal of maximizing happiness and minimizing suffering among those who can experience it. And specific facts about the structure of your soul can, sometimes, be evidence about your capability to experience suffering, despite not being the reason why the consideration is important in the first place. Though no one, to my knowledge, understands spiritology to a level deep enough to make evidence about suffering conclusive, as opposed to just suggestive.
And as mentioned before, getting consent is not the problem, even if of course harming those who consent is better than those who don't.

It seems they operate off different magical and metaphysical systems.  Eli suspected as much given the unprecedentedly unusual nature of the communication, but it’s still disappointing.  Still there are some points of similarity, even those few confused paragraphs have sparked a few idea for new experiments!

It sounds like we operate off of, and perhaps even exist within vastly different magical and metaphysical systems.  Still, in case any of this is useful…  [Several pages of magical and metaphysical theory, not exactly well organized, but better explained than Raz’s details about the soul.  A notable takeaway: the majority of souls in Eli’s universe are reincarnate, a small percentage are siphoned off by benevolent and malevolent extraplanar beings.  Also, the mentions of experimental evidence indicate Eli (or else he’s taking credit for other people’s work) has exhaustively and ruthlessly tested every aspect of the soul, reincarnation, and afterlife he can, with very little regard for suffering (some experiments put souls in near indefinite states of suffering or altered cognition, some utterly destroy souls, and some consign them to malevolent outsiders).]

But anyway, it seems to me “economic agent” and capacity for suffering and happiness are in practice near identical.  If a creature has a capacity to suffer or feel happy, it will nearly inevitably exert preferences for them, and thus act as an economic agent (if not a rational one).  Life is filled with both happiness and suffering, who are you to avoid suffering on behalf of another creature if it willingly consents?  Even if pursuing a value function of maximizing happiness and minimizing suffering across all creatures (as you seem to indicate, although I would encourage to consider if you truly care), an efficient economy is one of the best ways of achieving that goal!  And in the long run, scientific and magical progress are essential to achieving any goal as effectively as possible, so you shouldn’t let temporary smaller exchanges of happiness and suffering (so long as they are consensual) deter you from progress.

I did have a few questions [several pages of question on magic, focusing on souls and the afterlife].  On that note, you didn’t mention if you’ve discerned any afterlife within reach of your reality, many moral philosophers think quite a lot of suffering may be worth it for an eternally blissful afterlife.

Eli is starting to worry that this person is, despite their rational discourse and pragmatism (and not whining about Eli experimenting on babies), ultimately a moralist.  He’ll make one last entreaty and maybe try to elicit some more magical knowledge before giving up on the discussion.

Yeah, he gathered (from Buss, the reason he is even here, living this life) that the multi/meta/omniverse is big and has varying laws, often causing great linguistic complexity where the most accurate translation of a word refers to similar but very different concepts.
And some of the notes, if they are accurate, and can be trusted (both of which are doubtful, coming from the mad economist-necromancer), might be useful even if they are not immediately relevant.

My knowledge of the working of souls is mostly focused on interacting with the souls of still living creatures, usually my own; necromantic manipulation of bodies already dead and soulless; or necromantic transformation of souls in still living bodies to prevent them from experiencing the normal process occurring at the death of a body.
But, from what i do know:
On death, biochemical processes in the physical body stop. The connection between the soul and the body is severed. The vital energy in the soul dissipates, like water in a bucket full of holes, preventing the soul from having an anchor in the material world or ability to influence it; this might happen quickly or slowly, depending on the soul and the causes of death. The core of the soul survives, but it's hard to tell what the core actually is, or what it does. The astral body survives, but no longer reflecting a physical body, it can distort and take weird shapes. The mental/emotional part (sometimes called the fifth and sixth layers; this terminology seems inaccurate, because it is not clear whether they are two different parts, whether they are part of the core, and whether the seventh layer, which might or might not exist, also contains thoughts and emotions. Also the number counts the physical body as the first layer), might survive that process or not, or maybe survive in a diminished state.

Supposedly, after losing their anchoring on the material, souls move to the realms of gods.
I am not fully convinced gods exist, or that those realms exist. (Though if they do, it is possible that it happens because the core can be used as a source of spiritual energy (a plausible hypothesis, but one i am not sure about), and gods do use them for that purpose. All evidence of the contrary could be myths and rationalizations just as plausibly as being based on reality.)
I am sure that there is not any real evidence that souls retain the ability to act or make choices after death, and doubt but not fully sure that there is no evidence they even keep experiencing or feeling anything.

Because, importantly, there is a difference. Being able to act is not the same as being able to feel. Very small children are not able to act in a coherent way, not able to make plans or build models of the world, but they are able to feel and have preferences. That is the reason why simply having preferences is not the same as exerting preferences, and that being an economic agent is not the same as being a moral patient (though it largely is the same as being a moral agent, i think). And Economy is not the same as morality (though of course, a developed economy does usually help achieve many morally desirable goals).
I guess it's possible to imagine a scenario where i can't act on my own, like being physically paralysed, but i do think clearly and have coherent preferences and plans, and someone reads my mind, and acts based on what they see in my mind? And so i am technically an economic agent. But i don't think i would be exerting preferences in that case, as much as i would just have ideas, and someone else who has a preference to act on those ideas would exert their preference, while i will be fully dependant on them.

Finally, the question of consent.
Happiness is a result of alignment between a person's values and the events surrounding them. Suffering is a result of misalignment between a person's values and the events surrounding them. The values are "fulfilled" or "negated".
If we assume a very simple model, of only one relevant value, it is impossible to "willingly" suffer – if you will something to happen, it aligns with your values, and so you can't suffer from it.
In reality, most creatures have many values at once, and sometimes those come into conflict.
If i want to improve the world, and help my friends, and protect New Isengard, and also demonstrate my altruism and bravery and resilience, i will value giving my life force for the wraiths to drain. If i also want not to experience the feeling of all parts of my body, and some things that can't be part of my body but still feel that way, simultaneously go numb and cold and stretched out (which i am of course describing from experience), that means i also value that not happening! One value is fulfilled while the other is negated.
Just because i value doing that doesn't mean i wouldn't value not doing it even more. But then this is still a smaller net-loss of utility for me than it is for someone who doesn't want that to happen at all, and also hates the wraiths personally. So willingness/consent has value, but not enough of it to fully counteract the suffering. This is an important aspect of psychology and sociology that treating consent as a simple binary misses completely. (I predict a high probability that people, for all their lack of logic, would be more willing to listen to you if you actually understood the principles they base their lives, even if the things you are saying are the same things.)
If my goal is to reduce suffering (and yes that is my goal, i "truly care" without doubt, this is the only reason for my life), then consent solves part of the problem, but not all of it. A paper saying "i consent" doesn't solve the experience of suffering. Other people agreeing to do it isn't different from me agreeing to do it, it's still a problem for me, and reasonable to assume still a problem for them. If, as the first commenter suggested, i find someone who is genuinely unharmed by the experience itself, doesn't experience it as suffering, as opposed merely having additional reasons to value the action despite the suffering involved, that would solve the problem. But that would depend on the details of internal experience, not simple agreement or disagreement.

(For some dumb reason Raz felt previously that merely mentioning the existence of metamoral principles would automatically provide his collocutor with all the relevant information about that framework, like "several values can exist at once". This is an extremely obvious case of Illusory Transparency, mistake of a kind he never really makes.
Triple check the Communicator has no mental effects. Increase safety scans. Make larger pauses between communication sessions.
Stopping the usage still seems unjustified, considering the great benefit that was already provided.)

It does seem he is talking to a moralist, albeit a more pragmatic and rational one than usual.  But, the mention of Gods has sparked Eli's curiosity.  He'll ignore the increasingly pointless discussion about consent and focus on the fun parts like metaphysics!

You mentioned "Gods"... could you say more about that?

In my world, there has been several completely different things called "Gods" across various cultures and times in history:

Extraplanar beings, both malevolent and benevolent.  Some of them (usually the more malevolent and anarchic ones) have actively embraced this title and the varied cultural associations (such as supreme unbounded capabilities, innate archetypal natures, and some pattern of worship), while others have refused the label or at least clarified their limits (psychology they tend to be more "pure" than mortals in focus and goals and methods, and their magic seems of the same kind available to ordinary material plane beings, just aided by a few innate capabilities and the properties of their home planes).  Although they are extraplanar, they have discrete bodies, which can even be killed under the right circumstances.

The Archetypes fit various cultural associations of "God" much better, but in fact are limited by their very nature.  Their mind are embedded in conceptual space*, rendering them nearly utterly invulnerable and eternal with vast sensory percepts of reality related to their concepts.  Despite their vastness, they are quite limited in their ability to directly interact with reality: they each make possible a few unique spells, and have an ability to intervene through those spells, for example by weakening one of their spells that would oppose their interests or empowering one of their spells at an especially critical moment.

Various magical phenomena have been incorrectly attributed agency, but careful experimentation shows they respond to purely mechanical rules, without even the willful interventions of the Archetypes.

And finally, there are some entirely false stories about all-powerful being(s).  These stories persist through a variety of means, from deliberate social engineering (such as asserting moral value via the origin or enforcement of some powerful being), to the entertainment value of the stories, to the explanatory value of the story to ignorant people, to coincidental cultural association.  From your skepticism about your world's Gods, would you place them in this category?  Or do you have firmer evidence?

*(I am uncertain if your reality even has a conceptual space.  It is [several pages of dense jargon about how innate concepts interact with magic, Platonism, and more theoretical musing on nonmagical interaction with "pure concepts".])

I don't know much, is the whole thing.
Gods supposedly exist, and do things. But i haven't found any evidence yet that proves gods are the ones who do those things. For example, divine power is supposedly granted to priests as reward for their devotion, but, in my limited and subjective view, i don't expect anything to be different if divine power was actually a direct spiritual manifestation of the power of belief the priests themselves have. [Couple paragraph digression about soul layers, "Ba-Khion", spiritual energy, and the obscure unfalsifiable hypothesis about belief being a measurable form of energy.]
And gods never seem to be proactive, or make any decisions; people make those, using the gods as justification.
Spirits do in fact exist in immaterial planes, can be summoned, can grant favors or execute tasks, and make demands back of those who deal with them. I am not a shaman myself, but i do observe this happening often.
So maybe gods are in fact a very weird and powerful type of spirit. Or, precisely because spirits are a known thing, and their presence is visible and easily detected by those like me who knowledgeable of thaumatology, while gods aren't, that means gods don't really exist.
There are a lot of legends about gods doing things directly, sometimes incarnating in mortal bodies, but they are obscure, confused and ambiguous. Not actually proof, and barely evidence. "Would that rumor exist in a world where it wasn't actually true, and does that mean the thing id true if the rumor exists" is a hard question like that.

One historically fruitful line of inquiry from my world was an examination and exploration of the various taboos, heresies, and blasphemes of the faithful of various "Gods".  In the case of Archetypes, it allowed the determination that vast willful intelligences were behind certain spells in how they intervened to cut off power from heretics and blasphemers.  In the case of Extraplanar beings it revealed they lacked true omniscience to tracks blasphemes against themselves.  And in the case of other magical phenomena, it allowed the development of useful magics.  For instance, several cultures and religions had forbidden phrases they believed blasphemous to the gods, but where actually a naturally occurring discharge of magic in response to words.  The "forbidden" words simply discharged magical energy directly into the body of the speaker, killing them.  With a few modifications to the forbidden phrases, the discharge could be directed outsider their body, and thus the magic of Utterances was developed.

You've described your ethics around avoiding suffering?  Hopefully the hurt feelings of the religiously faithful (which are likely based on false premises anyway) aren't too big an obstacle to some scientific investigation?

And if they are an obstacle, Eli will write this person off as an incorrigible moralist and find a better way to spend his time.

The hurt feelings of people discovering a fact they believed to be false is the kind of thing that, while wouldn't exist in an ideal world, can't be avoided, and shouldn't be avoided. Doing so is never a good strategy that ends up in reduced suffering.
Most religious taboos i can think about are moral? Some of which i am already breaking. Necromancy, for example. I don't know of any cases like "saying this word will make the gods smite you". But they might exist, which might be useful for (carefuly) investigating the nature of gods, which is useful for the task of understanding reality, even if it's not a big priority.

Eli feels a flicker of frustration at the arbitrariness of what is deemed at “can’t be avoided”, but suppresses it.  Any acceptance of unavoidable suffering (as opposed to moralistic flailing) should be encouraged.

So it sounds like your world’s “Gods” are inactive enough there aren’t any useful starting points for scientific investigation?  Speaking of which… how far has science developed in your world?  I.e. have you reached the point of beginning to develop systematic reductionist mathematical models of the world?  It is speculated among my world’s philosophers that our scientific knowledge might reach such a point for “non-magical” science where a single unified mathematical model describes the underlying natures of diverse non-magical phenomena within the next few centuries.  For example, “magnetism” and “electricity” were recently shown to be interrelated and possible to create without magic.  And non-magical artificial chemical synthesis has recently produced several compounds previously thought to be strictly organic or magical.

Of course, I should clarify that one of the latest definitions defining magical versus non-magical explicitly references the extent to which phenomena reduce down to simple mathematics, so claiming we will understand all non-magical phenomena in a mathematical framework is somewhat circular.  And many phenomena classified as magic behave orderly enough that math describes them well, even if we can’t describe the phenomena in terms of simpler component elements.

Eli should have thought to probe for non-magical knowledge sooner, according to some metaphysical theories it is more likely to generalize and be reproducible across worlds, and thus there is more likely to have some idea Eli can profit from!  To be fair to himself, non-magical science really isn’t his area of expertise.

Not provably active enough. But that was already stated and doesn't need repeating.

Scientific investigation discovers many things! Not a full mode of the world, not even slightly, though i heard the claim that it must be possible. I don't know there to be a fundamental difference between magical and non-magical phenomena, merely phenomena caused by different forces.
But my knowledge, despite being large, is very limited, and maximally focused on practical concerns as opposed to general theories, mostly "how can a thing be used" rather than "why does it work that way and not another way". I am broadly familiar with the mechanism of electricity and magnetism, though.