# How can implement leap smearing in chrony?

**URL:** <https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119>\
**Category:** Uncategorized\
**Created:** [August 2, 2021, 3:41am UTC](https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119 "2021-08-02T03:41:17Z")\
**Posts on this page:** 5\
**Page:** 1

<div class="post-metadata">

**Author:** ![tennisbowling](https://avatars.discourse-cdn.com/v4/letter/t/71c47a/32.png) [@tennisbowling](https://community.ntppool.org/u/tennisbowling)\
**Post date:** [August 2, 2021, 3:41am UTC](https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119/1 "2021-08-02T03:41:17Z")

</div>

Hey all. I’m getting back into ntp, and I’m currently getting my time from gps/pps. How can implement leap smearing after the second/after?

---

<div class="post-metadata">

**Author:** ![alica](https://sea2.discourse-cdn.com/flex016/user_avatar/community.ntppool.org/alica/32/337_2.png) [@alica](https://community.ntppool.org/u/alica)\
**Post date:** [August 2, 2021, 11:53am UTC](https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119/2 "2021-08-02T11:53:33Z")

</div>

GPS by design does not implement smeared leap second; the PPS pulse will always adhere to the edge of 1 second. Leap smearing publishers like Facebook smeared the leap second on their secondary time servers before servicing outside world.

> **[Building a more accurate time service at Facebook scale](https://engineering.fb.com/2020/03/18/production-engineering/ntp-service/)**
>
> We've improved time precision within our infrastructure from 10 milliseconds to 100 microseconds, and released our own public NTP service.

If you are not willing to hack the code, you can synchronize your chrony against Google or Facebook directly.

---

<div class="post-metadata">

**Author:** ![tennisbowling](https://avatars.discourse-cdn.com/v4/letter/t/71c47a/32.png) [@tennisbowling](https://community.ntppool.org/u/tennisbowling)\
**Post date:** [August 2, 2021, 12:57pm UTC](https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119/3 "2021-08-02T12:57:29Z")

</div>

Ok, but if I leave it just on GPS/pps, it’s not even going to announce a leap second right?

---

<div class="post-metadata">

**Author:** ![mlichvar](https://avatars.discourse-cdn.com/v4/letter/m/e79b87/32.png) [@mlichvar](https://community.ntppool.org/u/mlichvar)\
**Post date:** [August 3, 2021, 9:06am UTC](https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119/4 "2021-08-03T09:06:09Z")

</div>

A leap smear should work even with refclocks, as long as they announce the leap second, or the leap second is available in the configured leapsectz zone.

Why do you need leap smear? The typical use case is a large company that has a mix of different NTP clients which cannot be configured to make a consistent slew correction. In smaller network it’s best to avoid it.

If you do enable leap smear on a server, please make sure it is not included in the pool.

---

<div class="post-metadata">

**Author:** ![Bas](https://sea2.discourse-cdn.com/flex016/user_avatar/community.ntppool.org/bas/32/465_2.png) [@Bas](https://community.ntppool.org/u/Bas)\
**Post date:** [August 3, 2021, 2:41pm UTC](https://community.ntppool.org/t/how-can-implement-leap-smearing-in-chrony/2119/5 "2021-08-03T14:41:46Z")

</div>

He’s using Google as NTP-source too, but that should not be done in pool-servers.

Stay away from Google if you use your server in the NTP-pool.