Something I forgot to mention earlier, pin-to-glass seals shouldn't leak due to thermal stress. The pins, if made correctly, have the same thermal expansion rate as the glass if it's made correctly.

They will leak some (hopefully) infinitesimal amount due to an imperfect seal, which is all but impossible to avoid.

I don't know what happened to your 6L6s. Mine are all RCAs, all at least 40 years old, and all fine, last time I checked.

I don't know that I've ever replaced a power amp tube except one that had a failed weld causing grids one and two to short. Actually, I've got another one with a failed weld. The getter fell off. I think it's in use, with the separated getter ring laying in just the right place. A little jarring could be disastrous.

It is not uncommon for power output tube to go gassy when
others seem to last a long time.
Dwight

That's usually due to the method of gettering, as far as I know. Many high power tubes only getter when the plate is hot (or is it some other element, filament? don't remember).

Most receiver tubes have getters that work when they are cold, ie, not in use. So they last longer when the equipment is turned off for long periods than high power transmitter tubes that only getter when hot.

I don't think there were 6L6s made that required power to getter. But, I suppose it's possible.

I can speak to this with some authority. All receiving tubes were gettered by being exposed to a strong RF field, which activated the getter. That this was so was evident by the silver coating at the top of the tube envelope. After that initial gettering that was it.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Yes, but that type of getter is always working, once activated. This can be seen very evidently when a tube envelope is compromised. It doesn't require to be heated at that moment to react with air.

An interesting device is the tunnel diode. They are negative resistance.
A latch can be made with two in series. One will be on the low resistance
part of the curve while the other is on the high resistance.
They are hard to get now days but one can make a device that is close to
the same and actually has better DC characteristics but won't run at as high
a frequency ( still oscillate at over 100MHz though ).
You put two junction FET in series, such that the gate of one goes to the drain
of the other, sources connected together. They need to be p-channel and
n-channel.
The curve is really nice. It starts at zero, goes to a peak and then all the
way back to zero.
Two of these ( 4 FETs ) in series makes a latch. Logic can be done as well.
One can make an entire computer with them.
Dwight
PS
I've made a simple oscillator with one of these, a tank circuit, a
bypass capacitor, a pot and a battery.
With the right tank, it will run at FM channel frequencies.
Unlike AM, if there is a stronger fixed frequency going into
the FM radio, the signal goes quiet.
Fun for jamming when you are listening to something with
friends and you say you don't like a particular song and
say, "I wish they'd turn that one off".
It then does just that. All the other stations work but that
one is quiet.
No one knows about the little box in your pocket with
the switch.

A tunnel diode (Esaki rectifier) is the poor man's Tetrode.

(Of course, the main use for an Esaki is its high speed, the negative resistance is a side benefit, just like it is with the Tetrode.)

Think about tunnel diodes next time you eat Pocky.