Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Anyone else having no email deliver with SendGrid?
98 points by samwillis on July 22, 2016 | hide | past | favorite | 80 comments
We have had no email delivered through SendGrid in over 16 hours. Their status says everything is normal but there are reports on Twitter of people having problems.

Is anyone else here having problems today?

SendGrid Support here! We wanted to step in and provide some color for what is going on. We’re currently making changes to our shared IP system in order for our customers to have a better delivery experience in the long term. As a result of this, the IP group you’re sending from has a new shared IP address(es).

As these IP groups warm up, you may see some deferrals if you are a Free or Essentials customer. However don't worry, this warm up period won't last long. Maybe a few days at most, or until major email receivers have enough data to determine the legitimacy of email being sent from these new IPs.

Keep in mind SendGrid will continue to attempt to deliver these throttled emails on your behalf for up to 72 hours (it rarely takes the full 72 hours to deliver an email throttled in this way).

If you wish to avoid disruptions like this in the future, considering upgrading your account to a Pro or higher plan (https://app.sendgrid.com/settings/billing), which includes a dedicated IP address as opposed to sending from a shared IP group. Dedicated IP addresses are great because instead of many different users sending from the same IP or group of IPs, you are in complete control of your sending reputation.

Customer feedback is extremely important us here at SendGrid, and we have made these changes as a result of that feedback. We know in the long run, this will immensely help your sending.

This will go away, just hang in there with us! If you have additional questions, please feel free to contact us by going to support.sendgrid.com.

We may not be a big customer to you and we may only send a few thousand transactional emails per month but even on the "essentials" paid package we expect a service level that is usable.

Suggesting that it's ok to wait a few days for our transactional email to deliver is frankly offensive. By definition transactional email is expected to be delivered in minutes if not seconds - days? That's got to be a joke?

This response is enough in its own to tip us into make a switch to an alternative supplier. It may be of no interest to SendGrid if an "essentials" customer leaves but if this is how you treat them you may find that more than a few do.

Frankly this entire situation has been mismanaged. As email "experts" you should have been able to migrate you customers over to the new IPs one by one as the IPs send level increases.

Finally suggesting that as this only effects some of your customers and so doesn't warrant being on the status page is just you making excuses to lie to your customers and hide the fact you have a serious issue. If this was a planed move with any possibility of delay you should have notified the effected customers before hand. We only discovered the problem when our customers stopped receiving emails and our support team saw an increase in support requests.

P.S. Your competitor Postmark has a very different oppinion about dedicated IPs https://postmarkapp.com/why/delivery

Disclosure: I am a former employee of a Sendgrid competitor.

The proper way to transition IPs is to split traffic between them, so when one IP is throttled, traffic can continue. These accounts should have been assigned the new IP in addition to the old one. Once warming completed after a couple of days, the old removed. I would strongly recommend all affected customers to open support tickets with Sendgrid to get put on good IPs.

> Suggesting that it's ok to wait a few days for our transactional email to deliver is frankly offensive

Another SendGrid customer here, in exactly the same situation (low-volume, purely transactional, paid "Essential" plan). But even if I were sending marketing blasts, a few days delivery delay would be unacceptable. It's hard not to hear SendGrid's response as "we don't really support anything below the Pro plan".

SendGrid has put up a support article about this: https://sendgrid.com/docs/Classroom/Deliver/shared_ip_thrott...

FWIW, a helpful SendGrid chat agent did move us back to a warmed-up IP pool once I complained, and our mails are going through again.

[Edit: clarify we're on a paid plan -- just not the Pro one.]

Nice to hear they are trying to help some customers.

I opened a support ticket 12 hours ago and have just been told it will eventually be sorted.

Try the chat option -- it's more immediate. (Though the chat agent did mention they're pretty backed up.)

Maybe SendGrid suggests using a courier instead of email for fast and guaranteed delivery of messages.

My company has been scratching our heads for the last 12 hours trying to figure out what was going on while waiting for SendGrid support to email us back. Without hearing anything I finally found this thread through a Twitter search, thank goodness.

I'm astounded that an email company couldn't do us the decency to email their customers to tell us what's going on. We've wasted valuable time and lost business because of them.

We're switching over to Postmark now. This year has been a nightmare in transactional email land from Mandrill to SendGrid to now Postmark, ugh. Hoping Postmark can do both you and I well.

Wow, you done goofed here. Where to start?

> However don't worry, this warm up period won't last long.

How long is "not long"? Your customers can't fix support problems this causes by saying "Well golly gee our vendor says it'll be over in not long"--in fact, thanks to you, they may not be able to email their own users at all.

> Keep in mind SendGrid will continue to attempt to deliver these throttled emails on your behalf for up to 72 hours

After which time...what? You'll summarily delete the emails? Refund the customers for failing to do your jobs? What exactly?

> Customer feedback is extremely important us here at SendGrid, and we have made these changes as a result of that feedback.

Sure, but you didn't warn the effected customers. Some of whom were paying you.


You all fucked up here, and this is probably going to lose you a lot of accounts. Good work.

Phone support said emails sold probably start going out Monday. But maybe Wednesday. If we switched to a Pro account it'd still be a few days for that IP to 'warm up'. We had no choice but to switch to Mandrill.

I'm not a customer, but seriously, you could at least apologize for disrupting customer's service.

The tone of this message is pretty terrible.

I'm also not a customer.

However, this message warns me not to sign up unless for a Pro account (don't know what that means).

You are essentially telling your Free or Essentials customers to go and f*ck off. Not nice.

Any deliberate maliciousness on their part aside, at the very least they could have "Warmed-up" the new IP addresses before switching customers over. That's like moving your existing customers over to a new server and only then having them wait while you "install software". At the worst, it's malicious. And at the best, just plain incompetent.

They need to send legit emails from those IP addresses to warm them up. It's not like a car that you can warm up by idling and not moving.

Again, should they not be able to do that before using the customers as guinea pigs? You'd think that they'd have a large buffer of "untainted" IP's that they slowly cycle towards. I.e. they "slowly" warm them up with legitimate emails of their own, and then have them on hand for when they need to replace existing IPs for customers.

I agree with you.

The point I was trying to make is this: in the example you gave ("waiting to install software") or the one I gave (warming up a car on the driveway), you can get the system ready for action without creating any impact in a production environment.

The case of email servers is an unusual example where you have to do real world tests in order to make the system work. In most other situations, you can do everything before the rubber hits the road.

So, whilst this could have been handled much better, it's not a simple situation.

I'm not a SendGrid user, but I'm curious as to if this was announced or sent out as a notification in some manner before it happened.

tl;dr: don't use sendgrid. no it wasn't. i have 100s of messages stuck in my queue marked as 'deferred'. these are critical emails that affect us financially. and the response is to "wait a few days". laughable. even without this process, lots of messages get stuck for 10 minutes or more with no response from their support department as to why. And this is a PAID service. over to AWS i go. i don't need complex reporting or elegant UI. just send the bloody email.

If your email reliability is affecting you financially maybe you should pay for a pro account? I would NEVER use a "free" service for e-mail if I depended on it for $ This is a lesson well learned for you.

i am paying sendgrid. i'm not using the "free" service either. Pro is 8x the cost i'm paying now. mandrill was the same price and actually delivered email. go figure. and i'm equally pissed at the horrible level of customer service as i am at the quality of the actual service.

If you're sending financially critical email, you probably shouldn't be using the $10/month tier of an online service. Email is complicated, I feel like most of these types of services explain the tradeoffs of each price point fairly well.

sorry, but i've used all sorts of services from free to paid. this has been the worst experience of all of them. mandrill was best. i'm pissed at the crappy customer service as much as i am about them deferring email (i understand the complexities) and we cant afford $1000 per year on email.

It sounds like you can't afford to NOT spend $1000 per year on email.


No notification was sent to my client who is using them.

The response to my support ticket on the subject was "things are working".

We're heavy SendGrid users and this is basically the excuse we've been looking for to jump ship after several years with them.

SG is incompetent with regards to email delivery. This should not come as a surprise to anyone that uses their service.

It took a month for them to confirm what we told them: users will still receive previously-queued emails even after they unsubscribe (!!!), meaning it's no surprise that they will flag future messages they receive as spam. A month or so later, SG acknowledged the bug but stated they had no intention of fixing that behavior at this time!

Their _core_ selling point is email delivery with low rejection and you don't need to build it yourself. They totally failed there as users are rightly marking their messages as spam _and_ we are therefore required to write, run, and maintain our own mail queuing service to check outgoing emails against _their own unsubscribe list_ before forwarding the email to SG to deliver for us. At this point, we use them only for the metrics, which is just not worth supporting such a sleazy operation.

Sendgrid is overly complex for what I need it to do. It's bloated with features, and over-complication. Their API to build a simple email necessitates creating at least three different objects.

They have a whitelisting feature, in case you are using your own domain you need to go through whitelisting steps. You need to configure your domain appropriately. It goes on and on. Then there are multiple ways to go about all of those things.

If you don't do it correctly then sometimes it doesn't work or a high amount of email ends up in spam. By default they enable features which some countries tightly control like tracking which all needs to be tinkered with to get right.

Then once it's all working perfectly I have seen it take up to 10 minutes for an email to be actually sent. I went with Sendgrid because they billed themselves as easy to use. You might be better off using a SMTP server of your own. You've got to configure it anyway.

I didn't down vote you; your comment adds to the discussion. The annoying SPF/DKIM additions to your DNS zone file are necessary for the basics of delivery to the inbox. Gmail now displays a scary red unlocked padlock for unsigned/unverified email. Running an SMTP server is indeed the way the internet was meant to handle email. However, the other tenants in your /25 CIDR block will ruin your sending IP address's reputation and you will face delivery to Spam and throttling unless you stay on top of blacklist removal requests. There are plenty of use cases for using your own SMTP server for sending, but I think the labor cost of doing so far exceeds the prices of Sparkpost, Postmark, Mailgun, Sendgrid, or hell, even Mailchimp.

I'm not a Sendgrid customer but their "warm up" answer here is totally insufficient. It strikes me as an attempt to spin a major outage as standard operating procedure, and as a result I'm not going to use their product in the future.

What leads you to believe that their explanation is not acceptable? I'm not saying I don't believe you, but more information about why you hold this opinion would be interesting to me.

To accept their explanation would be to accept that they went into this maintenance either with the intention of causing a several day long service outage or without knowledge that they would cause a service outage (i.e. something happened that they didn't anticipate.) If it was intentional, Bob help anyone who trusts them with their business. If it was accidental/unanticipated, they need to own up to what went wrong and show how they're going to prevent it in the future.

Same here, all of our emails have been Deferred since yesterday. We have a $9.95/month Essentials account.

Talked to chat support and they moved us to a new IP group and emails started being Delivered, however this lasted only 10 minutes.

After that all of our emails started being Deferred again. Contacted chat support again and they moved us again to a new IP group.

Let's see how much it lasts this time.

Chat support said having all of our emails Deferred for days with none being Delivered is normal and expected. We send transactional emails so having our customers wait days for the emails is not feasible for us.

We had recently switched from Mandrill, we'll have to start searching for a new provider.

I can heartily recommend Postmark. Generous free credits, excellent API, and one of the best DMARC tools (also free, https://dmarc.postmarkapp.com)

We have gone from Mandrill to Sendgrid and back to Mandrill.

We switched over to sendgrid after being pissed of with Mandrill due to their sudden change but switching to sendgrid was a bad decision in hindsight. Mandrill was rock solid and we never had issues. Sendgrid continues to have deliverability issues every once in a while and not to mention the blacklisting of their shared IPs (which I understand is a common problem with all providers but never happened with Mandrill for our business).

We have switched back to Mandrill (yes, got a paying mailchimp account just to use mandrill).

Not only sendgrid's UI is confusing, they don't even show the actual email content in their dashboard.

They gave two months' notice of the change to having to have a paying Mailchimp accounts with Mandrill - do you regard that as sudden? They also sent quite regular reminders, too. (I will admit we ended up leaving it a bit until the last minute, but it wasn't particularly arduous, even under moderate pressure.)

We use AWS' SES on a few projects, and, while it seems to work effortlessly, I have absolutely no idea how to get any logs out of it (like, anything beyond the absolute number of emails sent), which is a fairly major limitation for me.

Two months is very sudden for any enterprise software to make a large change in payment and account system, in my experience.

Same here. Had a chat with the support team and they say they had to allocate a number of new IPs, since some of their existing IPs were blacklisted. Now they need to warm up the new ones first, which may take an undefined amount of time.

Support was very helpful and migrated our traffic to already warmed up IPs.

Same experience here - we were migrated without warning 3 days ago and the majority of outgoing messages were subsequently deferred for 12 hours or more. When we reported the issue we were basically told they would be delivered eventually. No offer to migrate our traffic or to compensate us in any way.

We filed a support ticket and also had our traffic moved to a new IP, but I had to write to their Twitter account for them to pay attention. I urge everybody with an Essentials plan to do this, because it seems that if you don't complain on any channel possible you'll stay on this situation until it passes which can be "a few days" as they say.

what does warm up mean in this context?

I'd assume its a process of slowly ramping up new IP addresses to send out emails. If you take an IP and instantly use it on a massive scale, I'm sure it triggers a lot of spam filters.

In case others are looking for alternatives to SendGrid, SparkPost has been amazing to work with and offers 100,000 free transactional emails per month: https://www.sparkpost.com/

Whatever you do, don't consider Mandrill/Mailchimp. What they pulled earlier this year makes this issue look like nothing.

If you're using Django, I created django-anymail[1] earlier this year to simplify switching between full-featured transactional ESPs. This was a direct response to Mandrill's (relatively) short notice of major changes -- I used to maintain a popular Mandrill-Django integration.

[1]: https://github.com/anymail/django-anymail

Is it this ?

https://news.ycombinator.com/item?id=11203056 (Mandrill's Betrayal)


Another suggestion on excellent personal experience: postmark


Status page seems so to suggest nothing is wrong... http://status.sendgrid.com/

Looks like they just added a notice at the top of the status page, including "We apologize for any inconvenience this has caused."

Status page is for outages & maintenance only. Since this is not applicable to all accounts and should only last at most a few days we have decided not to post to our status page.

You would have done better to remain silent instead of making such posts.

A email delivery service isn't able to send those, and it considers that it isn't worth posting on their status page!

Are these messages really from SendGrid support? Looks like a desperate bid to prevent any future takeover attempts.

Are these messages really from SendGrid support?

Haha that's a great idea for a "dirty tricks" campaign against a competitor. Every time they have an outage, a greenbean HN user starts posting less and less plausible excuses with the aim of pissing everyone off. Eventually the real CEO has to get on and say, "We don't know who this is but please ignore them but no I don't want to say anything about our outage." Repeat for added lulz...

What's extra ironic is that Sendgrid employees were gloating on HN and posting inaccurate information about competitors when the Mandrill TOS change was announced. I think they really need an enforced social media policy.

Overall, I think Sendgrid does a fine job and it's unlikely this mistake will be repeated. These gaffes don't represent the majority of the company.

> These gaffes don't represent the majority of the company.

Maybe a disgruntled support person then.

Any reputable software company would post a partial outage on their status page, even if it only lasts for a few minutes

For many users email is a business critical function, the reason we pay other people to do deal with it is because it's a tricky business critical function.

If using yourselves is going to result in problems like this then I fail to see what we'd be paying for.

I truly hope that an impact duration of more than a few days is not a requirement for updating your status page.

Or even the other supposed requirement "effects _all_ accounts". Wow!

Yep for 12-13 hours now. One thing I've noticed is that the CNAMEs I set up for DKIM etc are not resolvable in DNS - could this be related to the issue?

All emails are listed under 'Activity' as being deferred.

Waiting for AWS to approve my SES access so I can switch away.

Same here, we have no delivered messages since yesterday at 17.00 GMT-3. Contacted Sendgrid support about this because we're having a lot of registered users that cannot activate their accounts.

I am using SendGrid on one project. Another surprising thing is that nobody from SendGrid is responding to this thread to explain and rectify as I have seen other companies do.

Perhaps you didn't see their helpful post here [0].

[0] https://news.ycombinator.com/item?id=12144124

They wouldn't have seen it because it wasn't posted yet.

We were previously using SendGrid but noticed a lot of email ended up in the Junk folders.

Now we're using Postmark; much better service and no issues with deliverability.

Yup. Will be leaving SendGrid. This is appalling.

No emails going out over here either. Our customers are complaining. Might be time to switch back to MailChimp.

I used SendGrid years ago, and never want to go back to it.

Then I switched to AWS. Nobody got fired for using AWS S3.

You mean SES?

No he literally writes his all emails to AWS s3. Just dumps them in a file.

Hence, no one can get fired. How can you get fired, if your fired email is just dumped in an s3 file?

You may have gotten downvoted to shit, but I thought that was funny!

Humor usually costs karma on hacker news but I can rebuild it with a few "wow Ruby is slow" comments.

S3 == SSS == Simple SMTP Service?

Are you using their shared IPs or do you have a dedicated one?

We are experiencing this issue also. Contacted SendGrid support and had them switch our traffic to different IP's.

Wish they would have communicated this in advance, or even update their status page.

If this guy was Homer Simpson, the answer would be:

(read using Homer voice)

"Uhhh, I love donuts! The status page is used only when something wrong happens. In your case, we decide to give you an outage by purpose, without telling you - haha - duh.

(Where is my beer, Marge?)

Oh, the customer is still waiting for my answer:

Dooonn't wooorrryyyy so much, we always do this on low plan customers - you know, we need some dumb guys to warm up our new IP range - hohoho - meeeryyyy x-mas!"

We've banned this account for violating the HN guidelines. If you don't want it to be banned, you're welcome to email hn@ycombinator.com. We're happy to unban people when there's reason to believe that they'll only post civil and substantive comments in the future.

We detached this comment from https://news.ycombinator.com/item?id=12144124 and marked it off-topic.


We detached this subthread from https://news.ycombinator.com/item?id=12144082 and marked it off-topic.

While funny, you're just making up words Sendgrid didn't say.

-----BEGIN PGP PUBLIC KEY BLOCK----- Version: BCPG C# v1.6.1.0

SendGrid under cover agent here...

Don't believe the conspiracy theorists out there, that this wasn't a push to the PRO plan for free tier usage account holders.

The free tier customers don't actually earn any money for send-grid, so I wouldn't be surprised if this little stunt increased subscriptions for the pro plan!

Remember, this wasn't an outage for anyone other than free tier account holders and that's why it wasn't on the outage page!

SendGrid does want your business, as long as you don't spend too long on the free-tier plan and ultimately upgrade!

Keep on Trucking...

Secret Agent... out!


No, it's is an outage for paid accounts. It is just apparently that you need a dedicated IP account to have not been effected.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact